{ "exported_at": "2026-11-15 21:24:18", "count": 334, "records": [ { "id": 5, "title": "Fuzzy interpolation systems with knowledge extraction and adaptation", "title_ar": "", "title_en": "Fuzzy interpolation systems with knowledge extraction and adaptation", "author": "Jie Li", "supervisor": "", "degree_type": "phd", "degree_evidence": "نص السجل: «PhD»", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "expert-systems", "research_subdomain": "ai-cybersecurity", "categories": "expert-systems; ai-applications; ai-cybersecurity", "keywords": "", "abstract": "Fuzzy inference system provides an effective means for representing and processing vagueness and imprecision. Conventional fuzzy modelling requires either complete experts’ knowledge or given datasets to generate rule bases such that the input spaces can be fully covered. Although fuzzy interpolation enhances the power of conventional fuzzy inference approaches by addressing the problem of lack of knowledge represented in the rule bases, it is still difficult for real-world applications to obtain sufficient experts’ knowledge and\/or data to generate a sparse rule base to support fuzzy interpolation. Also, the generated rule bases are usually fixed and therefore cannot support dynamic situations. In addition, all existing fuzzy interpolation approaches were developed based on the Mamdani fuzzy model, which are not applicable for the TSK fuzzy model. It significantly restricts the applicability of the TSK fuzzy inference systems. This PhD work, in the first part, presents a novel fuzzy inference approach, termed “TSK+ fuzzy inference”, to address the issue of performing the TSK inference over sparse rule bases. The proposed TSK+ fuzzy inference approach extends the conventional TSK fuzzy inference by considering the degree of similarity between given inputs and corresponding rule antecedents instead of conventional overlapped match degree, which allows TSK inference to be performed over sparse rule bases, dense rule bases, and imbalanced rule bases. In order to support the proposed TSK+ inference approach, a data-driven rule base generation method is also presented in this work. In addition, the proposed TSK+ inference approach has been further extended to deal with interval type-2 fuzzy sets. The effectiveness of this system in enhancing the TSK fuzzy inference is demonstrated through two real-world applications: a network intrusion detection system, and a network quality of service management system. In addition, in the second part of this work, a new rule base generation and adaptation method is developed in order to relax the requirement of rule base generation, which allows the fuzzy rule base to be generated with minimal or even without a priori knowledge. The proposed method mimics the pedagogic approach of experiential learning, which achieves automatic rule base generation and adaptation by transferring the proceeding performance experiences when performing inferences. The proposed rule base generation and adaptation method has been evaluated by not only a mathematical model but also a well-known control problem, inverted pendulum. The experimental results show that the control system can generate an applicable rule base to support the system running, thus demonstrating the effectiveness of the proposed approach.", "source_url": "https:\/\/openalex.org\/W2955571173", "pdf_url": "", "doi": "", "handle": "", "repository_name": "", "source_record_id": "W2955571173", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 8, "title": "Fuzzy model rozhodování investora do fotovoltaických technologií v předprojekční fázi", "title_ar": "", "title_en": "Fuzzy model rozhodování investora do fotovoltaických technologií v předprojekční fázi", "author": "Michal Pavlíček", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Brno University of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Czech Republic", "governorate": "", "publication_year": 2016, "language": "en", "research_domain": "computer-vision", "research_subdomain": "", "categories": "computer-vision; expert-systems", "keywords": "", "abstract": "Dissertation deals with fuzzy knowledge base supporting investor’s investment decision into the photovoltaic technologies during pre-design phase, when the engineering solution is not known yet. Probably, majority of investors have particular image about the cost and risk coming from investment into the photovoltaic technologies. However, this image is limited to the restricted knowledge areas of each investor. During the period, when the investor is planning investment into the costly photovoltaic technologies at complex, vague and heavily qualified information about the conditions and risks of investment in the specific region, where the environment is constantly inconsistent and multidimensional, it is possible to solve this complex situation with fuzzy logic. This dissertation is focusing on creation of fuzzy knowledge base with selected installed projects in Europe since 2008 and its use with expert system. Furthermore, is covering the definition and description of the variables, which are included in the investor decision making process. The complete designed architecture of the fuzzy knowledge base is tuned and five projects with different size are tested. The fuzzy knowledge base consists of overall 24 variables and 187 statements. The fuzzy knowledge base is tested with 5 projects.", "source_url": "https:\/\/openalex.org\/W2304027425", "pdf_url": "https:\/\/dspace.vut.cz\/bitstreams\/6ac89afb-b256-4d26-ae45-9ad176f90977\/download", "doi": "", "handle": "", "repository_name": "Brno University of Technology Digital Library (Brno University of Technology)", "source_record_id": "W2304027425", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 9, "title": "PEMBANGUNAN SISTEM PAKAR UNTUK PREDIKSI DIAGNOSA PENYAKIT GANGGUAN TIDUR DENGAN METODE FUZZY RULE-BASED CLASIFICATION SYSTEMS", "title_ar": "", "title_en": "", "author": "Mila Pradini Okvitasari", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Universitas Pendidikan Indonesia", "university_source": "repository_host", "college": "", "department": "", "country": "Indonesia", "governorate": "", "publication_year": 2016, "language": "id", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; medical-ai; expert-systems; predictive-analytics; transformers", "keywords": "", "abstract": "Gangguan tidur merupakan kelainan yang dapat menyebabkan masalah pada pola tidur seseorang, baik karena tidak bisa tertidur, sering terbangun pada malam hari atau ketidakmampuan untuk kembali tidur setelah terbangun. Hal ini dapat menyebabkan terjadinya kecelakaan, serta penurunan produktivitas pada seseorang. Pada saat ini salah satu cara untuk menyelesaikan permasalahan tersebut dapat dilakukan dengan cara konsultasi terhadap dokter. Namun hal ini menjadi masalah jika saat berkonsultasi harus disertakan dengan tatap muka. Yang mana semakin hari seseorang akan semakin sibuk dengan urusannya dan akhirnya menganggap sepele hal ini. Hal tersebut yang melatarbelakangi dilakukannya penelitian ini yang bertujuan untuk membuat sistem diagnosa penyakit gangguan tidur dengan metode Fuzzy Rule-Based Classifications Systems (FRBCS). FRBCS adalah suatu metode yang berbasis fuzzy rule, dimana FRBCS terdiri dua tahapan yaitu konstruksi model dan prediksi. Didalam konstruksi model dibutuhkan knowledge yang terdiri dari rule based dan database. Knowledge tersebut didapatkan dari seorang pakar, yang mana pakar tersebut mendefiniskan hubungan antara gejala dengan jenis gangguan tidur. Sedangkan prediksi adalah proses untuk mengambil keputusan dengan tahapan yaitu fuzzyfication, inference, dan classification. Adapun implementasi platform dalam penelitian ini yaitu dengan menggabungkan PHP dengan bahasa pemrograman R menggunakan package R-shinny. Untuk validasi sistem yang telah dibuat, beberapa eksperimen dilakukan dengan menggunakan data yang diperoleh dari suatu Rumah Sakit Jiwa di Jawa Barat. Dari eksperimen yang dilakukan dengan melakukan pengujian terhadap 33 data testing didapat akurasi sebesar 84.85% dengan hasil rata-rata 0.0133 detik untuk prosesnya. ;--- Sleep disorder is a disorder that can cause problems in a person's sleep patterns, either because they can not fall asleep, frequent waking during the night or the inability to return to sleep after waking. This can cause accidents, and decreased productivity in person. At this time one of the ways to solve these problems can be done by consulting the doctor. But this becomes a problem if the time of consultation should be included with face to face. Which is getting day someone will be busy with his business and finally to disregard it. This is the background of this research which aims to make the system a diagnosis of sleep disorders with Fuzzy Rule-Based Classifications Systems (FRBCS). FRBCS is a fuzzy rule-based method, which consists of two phases, namely FRBCS construction models and predictions. In the construction of the model required knowledge that consists of rule-based and database. The knowledge obtained from an expert to define the relationship between the type of sleep disorder symptoms. While the prediction is the process for taking decisions with a step that is fuzzyfication, inference, and classification. The implementation of this research platform which combine PHP programming language R using R-shiny package. For validation system that has been built, several experiments were conducted using data obtained from a Psychiatric Hospital in West Java. From the experiments carried out with a test of 33 testing data is obtained an accuracy of 84.85% with an average yield of 0.0133 seconds to process.", "source_url": "https:\/\/openalex.org\/W2901676950", "pdf_url": "http:\/\/repository.upi.edu\/26509\/2\/S_KOM_1200730_Abstract.pdf", "doi": "", "handle": "", "repository_name": "Repository at Universitas Pendidikan Indonesia (Universitas Pendidikan Indonesia)", "source_record_id": "W2901676950", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 10, "title": "Fuzzy Rule-Based Systems: Design, Analysis, and Applications", "title_ar": "", "title_en": "Fuzzy Rule-Based Systems: Design, Analysis, and Applications", "author": "Jeremy Kerr-Wilson", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Alberta", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "responsible-ai", "research_subdomain": "xai", "categories": "machine-learning; responsible-ai; expert-systems; data-mining; predictive-analytics; xai", "keywords": "", "abstract": "The extraction of knowledge from data is a relatively recent computational pursuit which has been the focus of significant research attention and has an extensive field of potential applications. With the advent of widespread data collection describing a variety of systems spanning many fields of expertise, the extraction of useful knowledge from data, for prediction or insight, has extraordinary potential and significant value. This realization has led to the development of machine learning, a collection of computational processes and algorithms through which sophisticated computer models can be constructed on the basis of data. Human centric systems, and more specifically certain forms of fuzzy systems, are a notable subset of machine learning which focus on knowledge extraction using multi-valued logic and set theory. Fuzzy systems are particularly well-suited to human-centric modelling as fuzzy sets describe real world systems, as perceived by humans, much more accurately than binary models. Fuzzy rule-based models are a form of fuzzy model which is particularly well-suited to human centric tasks due to the high degree of readability and interpretability conveyed to an expert reader. This, combined with their strong predictive ability, makes fuzzy rule-based systems an excellent candidate for those computational modelling pursuits where predictive accuracy may not be the singular requirement of a model. The objective of this dissertation is to design, analyze, and develop novel applications, methodologies, and algorithms for use with fuzzy rule-based systems, seeking to further their utility in predictive and human-centric modelling. In this dissertation, fuzzy rule-based systems are applied to different problem types, combined with existing computational data-structures and architectures, extracted from data in novel manners and formats, and analyzed to assess certain aspects of rule quality. Acknowledging the critical role of human centricity in computational modelling, we develop a set of fuzzy rule stability criteria which aim to quantify aspects of fuzzy rule quality while capturing critical non-numerical aspects of rule quality such as repeatability, consistency, and generalizability. We examine the generation of fuzzy rule-based models using hierarchical clustering and extract granular fuzzy models from data, forming information granules in the consequent parts of the fuzzy rules. We make use of fuzzy rule-based systems as the component models (weak learners) of a boosted ensemble, exploring their predictive power and adaptability in this environment. Finally, a novel fuzzy rule architecture is proposed using a hierarchical structure, alongside a generation procedure for extracting this hierarchical structure form data, with the aim of improving predictive performance and increasing the interpretability of the system.Each of these topics are justified with extensive experimental studies, using real-world data sets available from public repositories, demonstrating the feasibility or superiority of the proposed methods as compared to existing methodologies.", "source_url": "https:\/\/openalex.org\/W3163827966", "pdf_url": "", "doi": "10.7939\/r3-rmnz-bn16", "handle": "", "repository_name": "ERA: Education and Research Archive (University of Alberta)", "source_record_id": "W3163827966", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 15, "title": "An Expert System Integrated with a Bridge Information Management System (BrIMS), Cost Estimating, Deterioration Forecasting, and Linear Scheduling at the Conceptual Design Stage", "title_ar": "", "title_en": "An Expert System Integrated with a Bridge Information Management System (BrIMS), Cost Estimating, Deterioration Forecasting, and Linear Scheduling at the Conceptual Design Stage", "author": "Nizar Markiz", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Ottawa", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "machine-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; expert-systems; predictive-analytics", "keywords": "", "abstract": "Major bridge stakeholders such as federal and provincial transportation agencies are in dire need for objective knowledge-based systems that assist decision-makers in the selection of bridge type. Besides that, estimating bridge construction costs at the conceptual design stage is an increasing necessity for accurate budgeting and effective allocation of funding. Whilst multiple bridge management systems have already been developed; they still possess major drawbacks pertaining to interoperability and integration with complex time and cost optimization-related problem solving. In another perspective, infrastructure restoration has been backlogged with multifaceted factors that have captured the attention of municipal and federal authorities. Several successful integrations of bridge information management systems (BrIMS) with decision support systems and computer-aided engineering design solutions have significantly leveraged downstream processes of bridge maintenance operations and inspired many researchers. The subjective nature of evaluating bridge conditions and deteriorations is the main factor that influences bridge maintenance, repair, and replacement decisions. In order to overcome this shortcoming, the objectives of this study are intended to demonstrate the viability of integrating a decision support system with a stochastic gamma deterioration model utilizing a probabilistic fuzzy logic strategic approach at the conceptual design stage. In summary, this study presents a systematic multi-objective knowledge-based approach for selecting bridge type, forecasting elemental deteriorations, linear scheduling, and estimating construction costs at the conceptual design stage. The proposed methodology comprises a framework to deploy a system that automatically generates conceptual cost estimates by integrating objective functions with bridge information modeling (BrIM) through an external data interchange protocol in synchrony with interoperability standards. Deployment of the developed system shall minimize the degree of subjectivity involved while decision makings pertaining to bridge projects and assists designers and cost engineers obtain results in an integrated quantitative, qualitative, and systematic manner. The successful deployment of the expert system signifies a technological achievement of novelty pertaining to the integration of bridge information modeling (BrIM) concept with probabilistic fuzzy logic strategic approaches at the conceptual design stage of bridges.", "source_url": "https:\/\/openalex.org\/W2784626353", "pdf_url": "", "doi": "10.20381\/ruor-21467", "handle": "", "repository_name": "uO Research (University of Ottawa)", "source_record_id": "W2784626353", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 26, "title": "Driven segmentation of synthetic microwave images for breast cancer detection", "title_ar": "", "title_en": "Driven segmentation of synthetic microwave images for breast cancer detection", "author": "Bahar, Özlem", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Istanbul Technical University", "university_source": "repository_host", "college": "", "department": "", "country": "Turkey", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; medical-ai; image-segmentation; medical-imaging", "keywords": "", "abstract": "Breast cancer remains one of the most prevalent and life-threatening diseases worldwide, continuing to be the leading cause of cancer-related mortality among women. Current global health statistics reveal that millions of new breast cancer cases are diagnosed annually, with a disproportionate concentration in low- and middle-income countries (LMICs). In these regions, disparities in healthcare infrastructure, socioeconomic challenges, and limited access to advanced diagnostic technologies substantially hinder early detection efforts. Early diagnosis of breast cancer is of paramount importance, as timely intervention significantly improves patient survival rates, reduces the need for aggressive treatments, mitigates treatment-related side effects, and ultimately decreases the economic burden on healthcare systems. Despite technological advancements, conventional breast imaging modalities such as mammography, ultrasound, and magnetic resonance imaging (MRI) still face notable limitations globally in terms of accessibility, cost, and diagnostic efficacy. Mammography, regarded as the clinical gold standard for breast cancer screening, involves ionizing radiation and poses potential health risks with frequent use. Moreover, its diagnostic sensitivity is considerably reduced in women with dense breast tissue, leading to false-negative results and delayed treatment. Radiation-free ultrasound imaging requires highly skilled operators, resulting in variability in diagnostic accuracy and limiting its widespread application. Nuclear imaging techniques such as positron emission tomography (PET) and single-photon emission computed tomography (SPECT) provide valuable dynamic and functional imaging data; however, their high costs, radiation exposure, and limited availability restrict their broader clinical use. MRI offers superior soft tissue contrast and spatial resolution but is associated with high operational costs, infrastructure demands, and limited accessibility, especially in LMICs. Collectively, these challenges underscore the urgent need for safe, cost-effective, scalable, and reliable novel imaging technologies for early tumor detection. Microwave Imaging (MWI) has recently emerged as a promising alternative modality that enables non-ionizing electromagnetic wave-based examination of the internal structural and functional properties of breast tissue. The fundamental principle of MWI relies on the dielectric property differences—specifically dielectric permittivity (∈_r) and electrical conductivity (σ)—between malignant tumors and normal breast tissue. This contrast provides functional imaging information complementary to the anatomical data obtained from conventional methods. Microwave imaging systems primarily consist of transmitting and receiving antennas, with the collected data computationally reconstructed into images. Additionally, MWI systems are inherently more affordable, portable, and easier to operate, making them highly suitable for resource-constrained clinical environments. Nevertheless, the clinical translation of MWI has been impeded by intrinsic limitations such as relatively low spatial resolution, sensitivity to noise and artifacts, and challenges in accurately localizing tumors within reconstructed images. In response to these challenges, the present thesis proposes a comprehensive deep learning framework aimed at enhancing the diagnostic performance and tumor localization accuracy of microwave breast imaging. Within this scope, synthetic breast tissue models were developed using a high-fidelity simulation platform based on the Reverse Time Migration (RTM) algorithm, which realistically models the propagation and scattering of electromagnetic waves in heterogeneous and homogeneous biological tissues. The RTM algorithm processes source and receiver wavefields bidirectionally and establishes correlations between them to generate an image. It consists of two steps involving the backpropagation of complex conjugate data into the imaging domain and computing the cross-correlation norm of the forward and backward propagated fields. Synthetic breast models were generated employing microwave difference imaging based on the total electric field, necessitating knowledge of the healthy breast state during imaging. The total electric field data required for microwave imaging were synthetically produced via numerical solutions of electromagnetic scattering problems. This study investigates deep learning-based segmentation of low-resolution microwave images generated from total electric field data. The dataset comprises 1000 synthetic transverse breast images, each containing a single tumor, initially created at a high resolution of 400×400 pixels.", "source_url": "https:\/\/openalex.org\/W7112466934", "pdf_url": "https:\/\/polen.itu.edu.tr\/bitstreams\/45f1e73e-aeb2-438f-96e9-0efeb0835bfb\/download", "doi": "", "handle": "", "repository_name": "Istanbul Technical University Academic Open Archive (Istanbul Technical University)", "source_record_id": "W7112466934", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 30, "title": "Computer aided detection for breast lesion in ultrasound and mammography", "title_ar": "", "title_en": "Computer aided detection for breast lesion in ultrasound and mammography", "author": "Richa Agarwal", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; transfer-learning; cnn; image-segmentation; medical-imaging", "keywords": "", "abstract": "In the field of breast cancer imaging, traditional Computer Aided Detection (CAD) systems were designed using limited computing resources and used scanned films (poor image quality), resulting in less robust application process. Currently, with the advancements in technologies, it is possible to perform 3D imaging and also acquire high quality Full-Field Digital Mammogram (FFDM). Automated Breast Ultrasound (ABUS) has been proposed to produce a full 3D scan of the breast automatically with reduced operator dependency. When using ABUS, lesion segmentation and tracking changes over time are challenging tasks, as the 3D nature of the images make the analysis difficult and tedious for radiologists. One of the goals of this thesis is to develop a framework for breast lesion segmentation in ABUS volumes. The 3D lesion volume in combination with texture and contour analysis, could provide valuable information to assist radiologists in the diagnosis. Although ABUS volumes are of great interest, x-ray mammography is still the gold standard imaging modality used for breast cancer screening due to its fast acquisition and cost-effectiveness. Moreover, with the advent of deep learning methods based on Convolutional Neural Network (CNN), the modern CAD Systems are able to learn automatically which imaging features are more relevant to perform a diagnosis, boosting the usefulness of these systems. One of the limitations of CNNs is that they require large training datasets, which are very limited in the field of medical imaging. In this thesis, the issue of limited amount of dataset is addressed using two strategies: (i) by using image patches as inputs rather than full sized image, and (ii) use the concept of transfer learning, in which the knowledge obtained by training for one task is used for another related task (also known as domain adaptation). In this regard, firstly the CNN trained on a very large dataset of natural images is adapted to classify between mass and non-mass image patches in the Screen-Film Mammogram (SFM), and secondly the newly trained CNN model is adapted to detect masses in FFDM. The prospects of using transfer learning between natural images and FFDM is also investigated. Two public datasets CBIS-DDSM and INbreast have been used for the purpose. In the final phase of research, a fully automatic mass detection framework is proposed which uses the whole mammogram as the input (instead of image patches) and provides the localisation of the lesion within this mammogram as the output. For this purpose, OPTIMAM Mammography Image Database (OMI-DB) is used. The results obtained as part of this thesis showed higher performances compared to state-of-the-art methods, indicating that the proposed methods and frameworks have the potential to be implemented within advanced CAD systems, which can be used by radiologists in the breast cancer screening", "source_url": "https:\/\/openalex.org\/W3154566879", "pdf_url": "", "doi": "", "handle": "", "repository_name": "LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas)", "source_record_id": "W3154566879", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 35, "title": "Mammography Classification and Nodule Detection using Deep Neural Networks", "title_ar": "", "title_en": "Mammography Classification and Nodule Detection using Deep Neural Networks", "author": "Fabian Leander Sinzinger", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "KTH Royal Institute of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; cnn; image-classification; medical-imaging", "keywords": "", "abstract": "Mammographic screenings are the most common modality for an early detection of breast cancer, but a robust annotation of the depicted breast tissue presents an ongoing challenge, even for well-experienced radiologists. Computer-aided diagnosis systems can support the human classification. Modern systems therefore often rely on deep-learning based methods. This thesis investigates the fully automatic on-image classification of mammograms into one of the classes benign, malignant (cancerous) or normal. In this context, we compare two different design paradigms, one straightforward end-to-end model with a more complex decomposition hierarchy. While the end-to-end model consists mainly of the deep-learning based classifier, the decomposition pipeline incorporates multiple stages i.e. a region of interest detection (realized as a fully convolutional architecture) followed by a classification stage. Contrary to initial expectations, the end-to-end classifier turned out to obtain a superior performance in terms of accuracy (end-to-end: 76.57 %, decomposition: 65.66 %, computed as mean over all three classes in a one vs. all evaluation) and an improved area under receiver operating characteristic-score. All discussed parametric models were trained from scratch without using pre-trained network weights. Therefore we discuss the choice of hyper-parameters, initialization, and choice of a feasible cost function. For a successful feature extraction, in the region of interest detection stage, the negative dice coefficient proved itself to be a more robust cost function than the also investigated sensitivity-specificity loss.", "source_url": "https:\/\/openalex.org\/W2977003717", "pdf_url": "", "doi": "", "handle": "", "repository_name": "KTH Publication Database DiVA (KTH Royal Institute of Technology)", "source_record_id": "W2977003717", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 36, "title": "Computer-aided Detection of Breast Cancer in Digital Tomosynthesis Imaging Using Deep and Multiple Instance Learning", "title_ar": "", "title_en": "Computer-aided Detection of Breast Cancer in Digital Tomosynthesis Imaging Using Deep and Multiple Instance Learning", "author": "Mina Yousefi", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Concordia University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; generative-ai; medical-ai; cnn; medical-imaging", "keywords": "", "abstract": "Breast cancer is the most common cancer among women in the world. Nevertheless, early detection of breast cancer improves the chance of successful treatment. Digital breast tomosynthesis (DBT) as a new tomographic technique was developed to minimize the limitations of conventional digital mammography screening. A DBT is a quasi-three-dimensional image that is reconstructed from a small number of two-dimensional (2D) low-dose X-ray images. The 2D X-ray images are acquired over a limited angular around the breast. Our research aims to introduce computer-aided detection (CAD) frameworks to detect early signs of breast cancer in DBTs. In this thesis, we propose three CAD frameworks for detection of breast cancer in DBTs. The first CAD framework is based on hand-crafted feature extraction. Concerning early signs of breast cancer: mass, micro-calcifications, and bilateral asymmetry between left and right breast, the system includes three separate channels to detect each sign. Next two CAD frameworks automatically learn complex patterns of 2D slices using the deep convolutional neural network and the deep cardinality-restricted Boltzmann machines. Finally, the CAD frameworks employ a multiple-instance learning approach with randomized trees algorithm to classify DBT images based on extracted information from 2D slices. The frameworks operate on 2D slices which are generated from DBT volumes. These frameworks are developed and evaluated using 5,040 2D image slices obtained from 87 DBT volumes. We demonstrate the validation and usefulness of the proposed CAD frameworks within empirical experiments for detecting breast cancer in DBTs.", "source_url": "https:\/\/openalex.org\/W2904137965", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Spectrum Research Repository (Concordia University)", "source_record_id": "W2904137965", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 58, "title": "Predicting Irregularities in Arrival Times for Toronto Transit Buses with LSTM Recurrent Neural Networks Using Vehicle Locations and Weather Data", "title_ar": "", "title_en": "Predicting Irregularities in Arrival Times for Toronto Transit Buses with LSTM Recurrent Neural Networks Using Vehicle Locations and Weather Data", "author": "Anshuman Kush", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Trent University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "deep-learning", "research_subdomain": "rnn", "categories": "machine-learning; deep-learning; predictive-analytics; rnn", "keywords": "ANN, LSTM, Machine Learning", "abstract": "Public transportation systems play important role in the quality of life of citizens in any metropolitan city. However, public transportation authorities face criticisms from commuters due to irregularities in bus arrival times. For example, transit bus users often complain when they miss the bus because it arrived too early or too late at the bus stop. Due to these irregularities, commuters may miss important appointments, wait for too long at the bus stop, or arrive late for work. This thesis seeks to predict the occurrence of irregularities in bus arrival times by developing machine learning models that use GPS locations of transit buses provided by the Toronto Transit Commission (TTC) and hourly weather data. We found that in nearly 37% of the time, buses either arrive early or late by more than 5 minutes, suggesting room for improvement in the current strategies employed by transit authorities. We compared the performance of three machine learning models, for which our Long Short-Term Memory (LSTM) [13] model outperformed all other models in terms of accuracy. The error rate for LSTM model was the lowest among Artificial Neural Network (ANN) and support vector regression (SVR). The improved accuracy achieved by LSTM is due to its ability to adjust and update the weights of neurons while maintaining long-term dependencies when encountering new stream of data. Author Keywords: ANN, LSTM, Machine Learning", "source_url": "https:\/\/openalex.org\/W7207339572", "pdf_url": "https:\/\/trentu.scholaris.ca\/bitstreams\/9ca192dc-5548-476b-9d72-f695657df587\/download", "doi": "", "handle": "", "repository_name": "Digital Collections (Trent University)", "source_record_id": "W7207339572", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 65, "title": "Recurrent neural networks for time series prediction", "title_ar": "", "title_en": "Recurrent neural networks for time series prediction", "author": "Bernardo Perez Orozco", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Oxford", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "medical-ai", "research_subdomain": "rnn", "categories": "machine-learning; deep-learning; medical-ai; ai-applications; predictive-analytics; rnn; ai-smart-cities", "keywords": "", "abstract": "Attempting to predict the future long precedes the time where we could first quantify much of our present. But nowadays, performing long-term and reliable time series forecasting remains as relevant as ever across a vast number of application domains, including astronomy, healthcare, meteorology, physiology, energy systems, econometrics, finance and sociology, amongst many others. Practitioners act upon these forecasts, and it is thus mandatory that such forecasts are accompanied by well-calibrated uncertainty assessments so as to enable a better-informed decision-making process. This thesis proposes a novel time series forecasting framework, namely a scalable and general-purpose recurrent neural network approach which provides probabilistic predictions. Our method recasts time series forecasting as a symbolic sequence learning task by introducing a discrete encoding scheme over measurements. Such symbol sequence tasks have been most successfully dealt with by Long Short-Term Memory (LSTM) models and extensions, such as the sequence-to-sequence model, which form the core of our methodology. This presentation is accompanied by a large-scale experiment where we show the effectiveness of our method over 45 different datasets in comparison with state-of-the-art baselines, such as Gaussian Process Regression (GPR). Crucially, our framework also offers a number of extensions that address a broad range of use cases beyond univariate time series prediction. For example, we show how predictive densities over the occurrence of critical events within a predictive horizon can be inferred from our forecasts. Whereas this requires access to labelled data, we provide a complementary method for recognising and labelling prototypical time series segments on the basis of the feature embedding learnt by our neural networks. Such motifs can subsequently be used to build predictive densities and characterise time series. Lastly, we extend our framework to allow for multivariate forecasting, enabling efficient inference of joint predictive densities over multiple correlated time series. Finally, time series data scarcity constrains the choice of a forecasting model, especially as over-parameterised models such as recurrent neural networks cannot generalise from limited data. In this thesis, we propose a novel approach to infer accurate and reliable predictive distributions from scarce time series data. We achieve this through a recurrent neural network approach that infers a joint feature representation over the space of quantised time series by means of a compilation of auxiliary datasets. Cross-task knowledge transfer is readily enabled by assuming an ordinal regression approach, as quantised time series can then be characterised in terms of their ordinal bin sequences.", "source_url": "https:\/\/openalex.org\/W7116933523", "pdf_url": "https:\/\/ora.ox.ac.uk\/objects\/uuid:fb471520-8541-4abb-95a0-7649809cac87\/files\/mce9a2419de2282d545b0853644f60f4c", "doi": "10.5287\/ora-eoxk5gbqv", "handle": "", "repository_name": "Oxford University Research Archive (ORA) (University of Oxford)", "source_record_id": "W7116933523", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 66, "title": "Deep Learning for Medical Image Segmentation", "title_ar": "", "title_en": "Deep Learning for Medical Image Segmentation", "author": "Binyan Hu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; responsible-ai; ai-applications; self-semi-supervised; image-segmentation; medical-imaging; federated-learning; fairness; ai-cybersecurity", "keywords": "", "abstract": "Medical image segmentation (MIS) plays a fundamental role in a wide range of medical image analysis applications. Over the past decade, MIS has been significantly advanced by deep learning (DL). However, the effectiveness of DL-based methods typically depends on large-scale, precisely annotated datasets, which are often unavailable in the medical domain due to the high cost and expertise required for medical data acquisition and annotation. When trained on limited data, deep neural networks (DNNs) are prone to over-fitting, resulting in poor generalisation performance when deployed on previously unseen data. In MIS, this generalisation gap is exacerbated by two key challenges: 1) data scarcity: due to the limited quantity of annotated images available for training, over-fitting the training data causes poor generalisation performance on images unseen during training; 2) expert-specific annotation: variability in annotations across experts arises from the inherent complexity of medical images and differences in expert knowledge. Consequently, a model trained on certain experts' annotations may under-perform when deployed in clinical contexts with new images and new experts. This thesis aims to address these domain-specific limitations, with an overarching goal of generating enhanced supervision signals to improve the generalisation capability of DNNs for MIS across both images and user experts under constrained data settings. Toward achieving this goal, our contributions are threefold.First, we tackle the expert-specific annotation problem by leveraging the supervision from multiple experts via multi-expert annotation, aiming to achieve adaptive MIS that generalises to new experts' annotations. Through extensive empirical analysis, we investigate when and how multi-expert annotation improves a DNN model's generalisation capability to new experts' annotations and what DNN training method can best achieve such generalisation. Our findings provide detailed evidence for the efficacy of multi-expert annotation and practical insights into developing generalisable models that can efficiently adapt to new experts in the presence of expert-specific annotations.Second, we intend to mitigate the impact of data scarcity and improve the model's generalisation to new images by leveraging self-supervision.We develop an adaptive training method, which generates diverse self-supervised tasks from the existing annotated dataset, dynamically selects the most beneficial combination of tasks, and adaptively transfers the knowledge learnt from them to enrich the segmentation model's knowledge.Compared to existing approaches, our method integrates more diverse self-supervised tasks and offers superior effectiveness and efficiency in exploiting their supervision signals to benefit segmentation performance.The third study moves beyond the single-site setting and generates enhanced supervision via multi-site collaboration, where the goal is to improve generalisation performance for participating sites while navigating strict privacy constraints and potential security threats arising from non-transparent local processes. We propose a new federated learning method, tailored for the multi-site MIS scenario, to enable privacy-preserving defence against poisoning attacks by sustaining robust performance and localising the compromised site. Compared to previous approaches, our method provides utility-based defence and diagnosis, does not depend on public datasets, and works for the setting with few participating sites. These features enable more practical and effective federated learning with enhanced privacy preservation and security for multi-site MIS.Extensive experiments across diverse datasets and clinical settings demonstrate the effectiveness of the proposed methods in improving DNN model generalisation for MIS tasks. Collectively, this thesis contributes practical methodologies to address domain-specific challenges in MIS and advances the practical deployment of DNNs in real-world clinical environments.", "source_url": "https:\/\/openalex.org\/W7161281791", "pdf_url": "", "doi": "10.25916\/sut.32150593", "handle": "", "repository_name": "Swinburne figshare (Swinburne University of Technology)", "source_record_id": "W7161281791", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 67, "title": "Deep-Learning-Based 3D Medical Image Segmentation and Registration", "title_ar": "", "title_en": "Deep-Learning-Based 3D Medical Image Segmentation and Registration", "author": "Hao Xu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "The University of Sydney", "university_source": "repository_host", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; few-zero-shot; self-semi-supervised; transformers; multimodal; image-segmentation; medical-imaging", "keywords": "", "abstract": "Understanding the importance of image segmentation and registration in clinical diagnosis and treatment is critical for advancing precision medicine. Medical image analysis involves complex data processing tasks, particularly in scenarios with limited annotated data and multi-modality feature inconsistencies. Recent deep learning-based methods have provided breakthroughs in image segmentation and registration, especially in modalities like MRI and CT. However, challenges such as one-shot learning, incremental learning, and efficient deformable registration still persist. This thesis introduces a series of innovative deep learning frameworks, focusing on three one-shot segmentation settings and three Segment Anything Model (SAM) -assisted registration methods, to address the issues of annotation scarcity and feature misalignment in medical imaging. For segmentation, we develop three methods for one-shot, class-incremental, and novel-class tract segmentation. These incorporate uncertainty estimation, knowledge distillation, and voxel-level contrastive learning to enhance performance and generalization in one-shot settings. For registration, we propose three SAM-guided frameworks that leverage segmentation masks and anatomical prompts, as well as attention mechanisms, prototype learning, and contour-aware losses. These approaches achieve state-of-the-art accuracy in unsupervised deformable registration. Overall, our methods yield substantial improvements in segmentation and registration accuracy across diverse datasets and demonstrate strong potential for clinical application.", "source_url": "https:\/\/openalex.org\/W7126877936", "pdf_url": "", "doi": "", "handle": "", "repository_name": "The Sydney eScholarship Repository (The University of Sydney)", "source_record_id": "W7126877936", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 68, "title": "Efficient and Generalised Deep Learning Models for Medical Image Segmentation", "title_ar": "", "title_en": "Efficient and Generalised Deep Learning Models for Medical Image Segmentation", "author": "Zhendi Gong", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Nottingham", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; ai-applications; few-zero-shot; self-semi-supervised; cnn; transformers; image-segmentation; medical-imaging; ai-industry", "keywords": "", "abstract": "Medical image segmentation is a critical task in computer-assisted diagnosis and therapy planning. Accurate and efficient fully automated segmentation methods are essential for accelerating decision-making processes. Compared with natural image segmentation methods, current deep learning–based segmentation models for medical imaging face challenges related to data scarcity, inconsistent annotations, and high computational cost. Consequently, this thesis aims to develop efficiency and generalised medical image segmentation models, which are fundamental to the clinical adoption and industrial deployment of this technology in healthcare settings. This thesis systematically addresses these challenges through a series of progressively advanced deep learning models. The foundational work, CTranS, forms a baseline medical image segmentation model that combines a convolutional neural network (CNN) based encoder with a Transformer-based decoder. This initial contribution leverages the strengths of CNNs for local feature extraction and Transformers for global contextual understanding, achieving state-of-the-art (SOTA) performance. Subsequently, to enhance the generalisability, MO-CTranS is introduced, which is capable of learning from a mixture of multiple partially labelled datasets. MO-CTranS employs a CNN encoder and Transformer decoder with task-specific tokens to handle label inconsistencies and class imbalances, tackling the challenges of training a single unified model from fragmented and heterogeneously annotated medical data. This innovation moves beyond single-task, fully-annotated model training towards a more scalable and clinically realistic approach. Despite the superior performance of CTranS and MO-CTranS, the computational efficiency of the Transformer blocks hinders their clinical application. Thus, CRFTrans is proposed to enhance the model efficiency. As a novel recursive Vision Transformer layer designed to replace the commonly used cascaded Transformer layers, this approach represents a fundamental shift from simply stacking layers to a recursive refinement process, demonstrating that comparable performance can be achieved with significantly reduced computational complexity, thus facilitating deployment in resource-constrained clinical environments. However, these methods still depend on extensive manual annotation, which remains challenging and time-consuming in clinical practice. Thus, to further enhance both efficiency and generality, a novel semi-supervised medical image segmentation framework, referred to as SSL-MedSAM2, has been developed. SSL-MedSAM2 integrates a training-free few-shot segmentation branch for pseudo labelling and a fully supervised learning branch for label refinement. This design allows the model to leverage a small fraction of labelled images together with a large unlabelled set, dramatically decreasing the annotation effort. Together, these contributions result in deep learning models that are significantly more data-efficient and broadly applicable than conventional approaches. The overall impact of this work is to facilitate the reliable and scalable deployment of segmentation models in real-world clinical settings by bridging the gap among accuracy, efficiency, and generalisability.", "source_url": "https:\/\/openalex.org\/W7140609462", "pdf_url": "https:\/\/repository.nottingham.ac.uk\/bitstreams\/1a58d983-6aaa-4bbb-a40b-688e6d3af90c\/download", "doi": "10.17639\/7969", "handle": "", "repository_name": "Nottingham ePrints (University of Nottingham)", "source_record_id": "W7140609462", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 69, "title": "Efficient Deep Learning for Medical Imaging: Precision Segmentation and Beyond", "title_ar": "", "title_en": "Efficient Deep Learning for Medical Imaging: Precision Segmentation and Beyond", "author": "Lucas Gago", "supervisor": "", "degree_type": "phd", "degree_evidence": "اسم المستودع يخص أطاريح الدكتوراه: Tesis Doctorals en Xarxa (Consorci de Serveis Universitaris de Catalunya)", "university": "", "university_source": "", "college": "", "department": "", "country": "Spain", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; medical-ai; image-segmentation; medical-imaging", "keywords": "", "abstract": "[eng] This thesis advances automated medical image analysis by introducing four deep learning frameworks that systematically address core technical barriers to clinical deployment, including computational efficiency, variability in image quality, and the robust integration of imaging with clinical data. Through a compendium of research articles, we develop state-of-the-art solutions spanning ultrasound and MRI. First, we present an end-to-end framework for carotid intima-media thickness (CIMT) measurement in ultrasound images, achieving state-of-the-art atherosclerotic plaque characterization while delivering a 20x speed improvement (0.79 to 0.04 seconds per image). The system provides comprehensive outputs, including segmentation masks, automated measurements, and binary plaque detection, eliminating domain-specific post-processing requirements. Secondly, leveraging the features extracted by our end-to-end model, we pioneer their integration into clinical survival models, demonstrating that learned imaging biomarkers significantly enhance cardiovascular risk stratification with a 20% improvement in patient risk reclassification beyond traditional clinical variables. Third, we develop a multilevel EfficientNet-UNet++ architecture for 3D carotid vessel wall segmentation in black-blood MRI that achieves state-of-the-art performance through contextual slice concatenation and resolution optimization. The framework demonstrates optimal performance at 256 x 256 input resolution (6x original size) while maintaining computational efficiency through targeted multilevel processing. Finally, we introduce a quality-aware segmentation framework with custom loss functions for explicit quality modeling during training. When applied to ultrasound colon wall segmentation, this approach achieves a 20% improvement on medium-quality images and a 31% improvement on low-quality images, directly addressing ultrasound's fundamental challenge of variable image quality. Collectively, these contributions establish the technical foundations for robust clinical imaging through efficient segmentation architectures, the integration of imaging with clinical data, explicit quality modeling, and comprehensive clinical validation across two medical imaging modalities.", "source_url": "https:\/\/openalex.org\/W7146545742", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Tesis Doctorals en Xarxa (Consorci de Serveis Universitaris de Catalunya)", "source_record_id": "W7146545742", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 70, "title": "GENERALIZING MEDICAL IMAGE SEGMENTATION TASK WITH EFFICIENT DEEP LEARNING MODELS", "title_ar": "", "title_en": "GENERALIZING MEDICAL IMAGE SEGMENTATION TASK WITH EFFICIENT DEEP LEARNING MODELS", "author": "Abel A. Reyes-Angulo", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Michigan Technological University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; predictive-analytics; cnn; transformers; image-segmentation; medical-imaging", "keywords": "", "abstract": "Medical Image Segmentation is a critical task in the field of medical imaging, playing a crucial role in diagnostics, treatment planning, and disease monitoring. The emergence of Deep Learning (DL) has ushered in a new era in Artificial Intelligence (AI), propelling remarkable advancements in key domains like language translation, object recognition, and recommendation systems. This evolution has been accompanied by continuous enhancements in computational efficiency and improvements in predictive accuracy. The introduction of sophisticated algorithms, such as convolutional neural networks (CNNs) and transformers, exemplifies these advancements. DL algorithms have demonstrated exceptional efficacy in medical image segmentation tasks, showcasing the potential for AI-driven early diagnostics. However, the deployment of AI systems in clinical environments is often hindered by the substantial computational demands and complexity of cutting-edge DL models. In this research proposal, we explore various methodologies to enrich the visual feature representation for medical images. We focus on integrating global context-oriented techniques, such as attention mechanisms, into the development of parameter-efficient deep learning models. Our goal is to create a generalized, end-to-end medical image segmentation framework that can accurately and efficiently segment medical images across different modalities and conditions. By leveraging advanced deep learning techniques and optimizing model architectures, we aim to enhance the performance and generalization capabilities of medical image segmentation models, ultimately contributing to improved clinical outcomes", "source_url": "https:\/\/openalex.org\/W4410497124", "pdf_url": "", "doi": "10.37099\/mtu.dc.etdr\/1941", "handle": "", "repository_name": "Digital Commons - Michigan Tech (Michigan Technological University)", "source_record_id": "W4410497124", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 73, "title": "Active deep learning for medical imaging segmentation", "title_ar": "", "title_en": "Active deep learning for medical imaging segmentation", "author": "Marc Górriz Blanch", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Spain", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; predictive-analytics; cnn; image-classification; image-segmentation; medical-imaging", "keywords": "", "abstract": "This thesis proposes a novel active learning framework capable to train effectively a convolutional neural network for semantic segmentation of medical imaging, with a limited amount of training labeled data. Our approach tries to apply in segmentation existing active learning techniques, which is becoming an important topic today because of the many problems caused by the lack of large amounts of data. We explore different strategies to study the image information and introduce a previously used cost-effective active learning method based on the selection of high confidence predictions to assign automatically pseudo-labels with the aim of reducing the manual annotations. First, we made a simple application for handwritten digit classification to get started to the methodology and then we test the system with a medical image database for the treatment of melanoma skin cancer. Finally, we compared the traditional training methods with our active learning proposals, specifying the conditions and parameters required for it to be optimal.", "source_url": "https:\/\/openalex.org\/W2768586247", "pdf_url": "", "doi": "", "handle": "", "repository_name": "RECERCAT (Consorci de Serveis Universitaris de Catalunya)", "source_record_id": "W2768586247", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 78, "title": "Privacy-preserving and data-efficient deep learning for medical image segmentation", "title_ar": "", "title_en": "Privacy-preserving and data-efficient deep learning for medical image segmentation", "author": "Zhou ZHENG", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Japan", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; medical-ai; responsible-ai; image-segmentation; medical-imaging; fairness", "keywords": "", "abstract": "Medical image segmentation that aims to delineate regions of interest (ROIs) has been a long-standing problem in medical image analysis.It plays an essential role in computer-aided diagnosis (CAD) and computer-aided surgery (CAS).The development of Deep learning (DL) techniques has significantly advanced these areas by providing accurate and automatic segmentation.Despite such remarkable progress and achievements, DL in medical image segmentation still faces several challenges, typically privacy preservation and data efficiency.On one aspect, DL typically requires numerous samples for training.However, sharing medical images among hospitals or medical institutes is usually impeded due to privacy regulations, making it hard to collect large-scale datasets.Moreover, the performance of DL models greatly depends on fully and densely annotated labels, but obtaining such ideal annotations is a bottleneck, particularly in the medical field.By contrast, medical image datasets usually come with imperfect annotations in practice.These limitations emphasize the importance of developing privacy-preserving and data-efficient DL techniques for medical image segmentation with highly distributed datasets accompanying imperfect annotations.Driven by these perspectives, this thesis reevaluates the application of DL in medical image segmentation, particularly the 3-dimensional (3D) abdominal multi-organ segmentation from computed tomography (CT) images and the surgical instrument segi LIST OF FIGURES 4.5 Ablation study on different methods adopted by local clients for handling various imperfect annotations.IL: ideal annotation type.SL: scarce annotation type.NL: noisy annotation type.", "source_url": "https:\/\/openalex.org\/W7145109004", "pdf_url": "https:\/\/nagoya.repo.nii.ac.jp\/record\/2011569\/files\/k15072_thesis.pdf", "doi": "", "handle": "", "repository_name": "Institutional Repositories DataBase (IRDB)", "source_record_id": "W7145109004", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 81, "title": "Deep learning based segmentation for multi-modal medical MRI Images", "title_ar": "", "title_en": "", "author": "Mustafa Arikan", "supervisor": "", "degree_type": "master", "degree_evidence": "نص السجل: «Masterarbeit»", "university": "University of Vienna", "university_source": "repository_host", "college": "", "department": "", "country": "Austria", "governorate": "", "publication_year": 2018, "language": "de", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; medical-ai; multimodal; image-segmentation; medical-imaging", "keywords": "", "abstract": "In dieser Masterarbeit stellen wir einen automatischen Algorithmus für Segmentierung von multi-modalen Datensätzen mittels Deep Learning vor. Unser Algorithmus basiert auf der SegNet Architektur. Wir haben SegNet erwietert, sodass dieses nicht nur die drei Kanäle eines RGB-Bildes, sondern eine beliebige Anzahl von Kanälen von multi-modalen Bildern verarbeiten kann. Weiters haben wir Visualisierungen für das Untersuchen von genauen und robusten Modellen entwickelt. Wir evaluieren unseren Algorithmus auf manuell segmentierten MRI Datensätzen und vergleichen mit einem State-of-the-Art Algorithmus.", "source_url": "https:\/\/openalex.org\/W3157823582", "pdf_url": "", "doi": "10.25365\/thesis.55604", "handle": "", "repository_name": "University of Vienna", "source_record_id": "W3157823582", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 83, "title": "Volumetric medical image segmentation with deep learning pipelines", "title_ar": "", "title_en": "Volumetric medical image segmentation with deep learning pipelines", "author": "Imad Eddine Toubal", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Missouri", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; medical-ai; cnn; image-segmentation; medical-imaging", "keywords": "", "abstract": "Automated semantic segmentation in the domain of medical imaging can enable a faster, more reliable, and more affordable clinical workflow. Fully convolutional networks (FCNs) have been heavily used in this area due to the level of success that they have achieved. In this work, we first leverage recent architectural innovations to make an initial segmentation: (i) spatial and channel-wise squeeze and excitation mechanism; (ii) a 3D U-Net++ network and deep supervision. Second, we use classical methods for refining the initial segmentation: (i) spatial normalization and (ii) local 3D refinement network applied to patches. Finally, we put our methods together in a novel segmentation pipeline. We train and evaluate our models and pipelines on a dataset of a 120 abdominal magnetic resonance - volumetric - images (MRIs). The goal is to segment five different organs of interest (ORI): liver, kidneys, stomach, duodenum, and large bowel. Our experiments show that we can generate full resolution segmentation of comparable quality to the state-of-the-art methods without adding computational cost.", "source_url": "https:\/\/openalex.org\/W4321613625", "pdf_url": "https:\/\/mospace.umsystem.edu\/xmlui\/bitstream\/10355\/94345\/1\/ToubalImadEddineResearch.pdf", "doi": "10.32469\/10355\/94345", "handle": "", "repository_name": "MOspace Institutional Repository (University of Missouri)", "source_record_id": "W4321613625", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 84, "title": "Impact of Rater Style on Deep Learning Segmentation in Medical Imaging", "title_ar": "", "title_en": "", "author": "Olivier Vincent", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "École Polytechnique de Montréal", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2021, "language": "fr", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; medical-ai; responsible-ai; image-segmentation; medical-imaging; fairness", "keywords": "", "abstract": "La sclerose en plaques est la maladie auto-immune la plus courante du systeme nerveux central. Elle se caracterise par la presence de lesions dans le cerveau et la moelle epiniere, visibles en imagerie par resonance magnetique (IRM). Cependant, pour extraire des infor-mations utiles des images, il est necessaire de segmenter les lesions sur ces images, ce qui est un processus long et couteux lorsqu’il est realise manuellement par un expert tel qu’un radiologue. L’objectif de ce projet est d’utiliser des methodes innovantes d’apprentissage profond pour ameliorer la segmentation des images medicales. Premierement, pour aborder la generalisation a travers di˙erents contrastes dans la segmen-tation de la sclerose en plaques de la moelle epiniere, nous mettons en œuvre la modula-tion lineaire par caracteristique (FiLM) pour tirer parti de la grande variete de parametres d’acquisitions IRM dans le modele de segmentation, en apprenant les caracteristiques de chaque contraste. Fait interessant, un U-Net bien optimise a atteint les memes performances que notre FiLMed-Unet sur un ensemble de donnees a contrastes multiples (0,72 de score Dice), ce qui suggere qu’il existe un goulot d’etranglement dans cette tâche, qui n’est pas la generalisation a travers di˙erents contrastes. Ce goulot d’etranglement provient probable-ment de la variabilite interexperts, qui est estimee a 0,61 de score Dice dans notre ensemble de donnees. Deuxiemement, afin de s’attaquer a ce goulot d’etranglement, nous quantifions le style des experts qui annotent les donnees sous forme de biais et de consistance. Cela nous permet ensuite d’explorer l’impact des styles d’annotations sur les modeles d’apprentissage profond. Deux ensembles de donnees publics multievaluateurs et multicentriques sont utilises, un de lesions de sclerose en plaques cerebrales et un de segmentation de la matiere grise de la moelle epiniere. Sur les deux ensembles de donnees, les resultats montrent une correlation (R2 = 0, 60 et 0, 93) entre le biais de l’expert et l’incertitude du modele d’apprentissage profond. L’impact de la fusion d’annotations des experts sur cette relation est egalement etudie, et nous montrons que les consensus multicentriques sont plus eycaces que les con-sensus monocentriques pour reduire l’incertitude, car le style de l’expert n’est pas specifique a l’individu, mais principalement au centre.----------ABSTRACT Multiple sclerosis is the most common autoimmune disease of the central nervous system. It is characterized by the presence of lesions in the brain and spinal cord, which are visible in magnetic resonance imaging (MRI). However, to extract useful information from the images, it is necessary to segment the lesions on these images, which is a long and expensive process when performed manually by an expert such as a radiologist. The goal of this project is to use innovative deep learning methods to improve segmentation of medical images. First, to tackle generalization across imaging contrasts in spinal cord multiple sclerosis seg-mentation we implement Feature-wise Linear Modulation (FiLM) to leverage physics knowl-edge within the segmentation model and learn the characteristics of each contrast. Inter-estingly, a well-optimized U-Net reached the same performance as our FiLMed-Unet on a multi-contrast dataset (0.72 of Dice score), which suggests that there is a bottleneck in spinal MS lesion segmentation di˙erent from the generalization across varying contrasts. This bot-tleneck likely stems from inter-rater variability, which is estimated at 0.61 of Dice score in our dataset. Second, as a follow-up we quantify rater style in the form of bias and consistency and explore the impacts on deep learning models. Two multi-rater and multi-center public datasets are used, consisting of brain multiple sclerosis lesion and spinal cord grey matter segmentation. On both datasets, results show a correlation (R2 = 0.60 and 0.93) between rater bias and deep learning uncertainty. The impact of label fusion between raters’ annotations on this relationship is also explored, and we show that multi-center consensuses are more e˙ective than single-center consensuses to reduce uncertainty, since rater style is not individual-specific but mostly center-specific.", "source_url": "https:\/\/openalex.org\/W3211467155", "pdf_url": "https:\/\/publications.polymtl.ca\/6283\/1\/2021_OlivierVincent.pdf", "doi": "", "handle": "", "repository_name": "PolyPublie (École Polytechnique de Montréal)", "source_record_id": "W3211467155", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 85, "title": "Quantifying and understanding uncertainty in deep-learning-based medical image segmentation", "title_ar": "", "title_en": "", "author": "Benjamin Lambert", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2024, "language": "fr", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "computer-vision; medical-ai; image-segmentation; medical-imaging", "keywords": "", "abstract": "Quantification et caractérisation de l'incertitude de segmentation d'images médicales pardes réseaux profonds Ces dernières années, les algorithmes d'intelligence artificielle ont démontré des performances exceptionnelles dans de nombreuses tâches, incluant la segmentation et classification d'images médicales. La segmentation automatique des lésions dans des IRMs du cerveau permet une quantification rapide de la progression de la maladie : un compte des nouvelles lésions, une mesure du volume lésionnel total et une description de la forme des lésions. Cette analyse peut ensuite être exploitée par le neuro-radiologue qui peut s'en servir pour adapter le traitement thérapeutique si nécessaire. Cela permet de rendre la décision médicale plus rapide et plus précise.Actuellement ces algorithmes, que l'on considère souvent comme des boîtes noires, produisent des prédictions sans aucune information concernant leur certitude. Cela empêche la pleine adoption des algorithmes d'intelligence artificielle dans les domaines sensibles, car ils ont en effet tendance à produire des erreurs avec une grande confiance, pouvant tromper les décideurs humains. Identifier et comprendre les causes de ces échecs est un élément clé pour maximiser l'utilité des algorithmes d'intelligence artificielle et permettre leur acceptation au sein du corps médical. Pour atteindre ce but, il est important de pouvoir distinguer entre les deux principales sources d'incertitude. Tout d'abord, l'incertitude aléatoire, qui correspond à l'incertitude liée au bruit intrinsèque de l'image, et les artefacts d'acquisition. Ensuite, l'incertitude épistémique, qui a attrait au manque de connaissance du modèle.L'objectif conjoint de Pixyl et du GIN est d'atteindre une meilleure identification des sources d'incertitude dans les réseaux de neurones profonds, et par conséquent développer de nouvelles méthodes pour estimer cette incertitude dans le cadre d'une utilisation clinique routinière et en temps réel.Dans le cadre de la segmentation d’images médicales, l’estimation de l’incertitude est pertinente à plusieurs échelles. Tout d’abord, à l’échelle du voxel, l’incertitude peut être quantifiée par le biais de cartes d’incertitude. Cela permet de superposer l’image, sa segmentation et la carte d’incertitude pour visualiser les régions incertaines. Ensuite, pour les pathologies comme la Sclérose-en-Plaques, l’attention du radiologue se situe à l’échelle de la lésion plutôt que du voxel. L’estimation d’incertitude structurelle, c’est-à-dire à l’échelle de la lésion, permet au radiologue de rapidement contrôler les lésions incertaines qui peuvent être des faux positifs. Troisièmement, des métriques haut-niveau comme le volume ou le nombre de lésions sont couramment extraites des segmentations. Être capable d’associer à ces métriques des intervalles prédictifs est important pour que le clinicien puisse prendre en compte cette incertitude dans son analyse. Pour finir, l’incertain peut être quantifiée à l’échelle de l’image entière, par exemple pour détecter les images hors-distributions qui présentent une anomalie importante pouvant biaiser leur analyse.Dans cette thèse, des outils de quantification d’incertitude opérant à chacun de ces niveaux sont développés. Plus généralement, les méthodes désirées et attendues devront permettre à Pixyl d'améliorer ces modèles, services et produits actuels. Pour l'application clinique, le temps d'inférence est particulièrement critique : l'assistance à la prise de décision est utile seulement si elle est suffisamment rapide pour être appliquée pendant la consultation du patient (c'est-à-dire en moins de 5 minutes). De plus, les solutions innovantes devront maintenir un haut niveau de performance même appliquée sur de petites bases de données d'image, ce qui est généralement le cas dans le domaine médical.", "source_url": "https:\/\/openalex.org\/W7151869013", "pdf_url": "", "doi": "10.70675\/a043da9aze69dz4ea2z978dzcd9a570ce0a7", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7151869013", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 86, "title": "Medical image segmentation with belief function theory and deep learning", "title_ar": "", "title_en": "", "author": "Ling Huang", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2023, "language": "fr", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; medical-ai; multimodal; image-segmentation; medical-imaging", "keywords": "", "abstract": "Segmentation d'images médicales avec la théorie de la fonction de croyance et l'apprentissage en profondeur L'apprentissage profond a montré des contributions prometteuses dans la segmentation d'images médicales avec de puissantes capacités d'apprentissage et de représentation de caractéristiques. Cependant, il présente des limites pour le raisonnement avec et la combinaison d'informations imparfaites (imprécises, incertaines et partielles). Dans cette thèse, nous étudions les approches de segmentation d'images médicales avec la théorie des fonctions de croyance et l'apprentissage profond, en nous concentrant spécifiquement sur la modélisation et la fusion d'informations basées sur des preuves incertaines. Tout d'abord, nous passons en revue les méthodes de segmentation d'images médicales existantes basées sur la théorie de la fonction de croyance et discutons de leurs avantages et défis. Deuxièmement, nous présentons un cadre de segmentation d'images médicales semi-supervisé pour diminuer l'incertitude causée par le manque d'annotations avec la segmentation évidentielle et la fusion de preuves. Troisièmement, nous comparons deux classificateurs d'évidence, le réseau neuronal d'évidence et le réseau de fonction de base radiale, et nous montrons l'efficacité de la théorie de la fonction de croyance dans la quantification de l'incertitude ; nous utilisons les deux classificateurs d'évidence avec des réseaux neuronaux profonds pour construire des modèles d'évidence profonds pour la segmentation du lymphome. Quatrièmement, nous présentons un cadre de fusion d'images médicales multimodales prenant en compte la fiabilité de chaque source d'image RM lors de l'exécution de différentes tâches de segmentation à l'aide de fonctions de masse et d'actualisation contextuelle.", "source_url": "https:\/\/openalex.org\/W7151868992", "pdf_url": "", "doi": "10.70675\/8a8a9e2bz03f7z4d78z879cz9b80ab963e96", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7151868992", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 87, "title": "Deep learning methods for localization, segmentation and robustness in medical imaging", "title_ar": "", "title_en": "", "author": "Othmane Laousy", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2024, "language": "fr", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; medical-ai; reinforcement-learning; image-segmentation; medical-imaging", "keywords": "", "abstract": "Méthodes d'apprentissage profond pour la localisation, la segmentation et la robustesse en imagerie médicale Ces dernières années, nous avons assisté à un essor remarquable des avancées à la croisée de l'apprentissage profond et de la médecine, en particulier dans le domaine de l'imagerie médicale. Cette évolution rapide a engendré une multitude de défis nécessitant une attention particulière.L'un des principaux défis découlant de l'évolution des techniques d'apprentissage profond est la demande de données étendues et diversifiées pour l'entraînement des modèles. A mesure que les modèles deviennent plus sophistiqués, leur besoin en données s'intensifie, ce qui constitue un obstacle à l'acquisition d'ensembles de données non seulement volumineux, mais aussi véritablement représentatifs des divers scénarios médicaux rencontrés en pratique.Une autre préoccupation majeure concerne la vulnérabilité des modèles d'intelligence artificielle aux attaques adversariales, où de petites perturbations des données d'entrée peuvent entraîner des répercussions significatives sur la fiabilité du modèle. La sensibilité de ces modèles aux manipulations adverses constitue une menace potentielle pour leur déploiement dans des scénarios médicaux réels, où les conséquences de ces erreurs peuvent être désastreuses. Il est impératif de protéger les modèles d'IA contre les attaques adverses pour inspirer confiance aux professionnels de la santé et aux patients qui comptent sur ces technologies pour obtenir des diagnostics précis et des recommandations de traitement.Cette thèse explore ces défis pressants, offrant des perspectives sur les directions potentielles pour atténuer l'impact de la rareté des données et améliorer la fiabilité des modèles d'IA face aux attaques adversariales.Tout d'abord, nous explorons l'apprentissage par renforcement profond (Deep Reinforcement Learning, DRL) pour aborder les problèmes liés à l'apprentissage dans des contextes où les données sont rares. Cette thèse présente une application novatrice du DRL dans l'analyse d'images médicales, en se concentrant particulièrement sur les défis de localisation et de segmentation pour la quantification de la sarcopénie à partir de scans CT. En optimisant la prise de décision d'un agent par essais et erreurs, nous avons développé un réseau Deep Q-Network qui automatise l'identification de la coupe de la troisième vertèbre lombaire. Pour la segmentation, nous avons employé un algorithme d'optimisation de politique proximale pour tracer les contours des muscles psoas et parietaux, automatisant le calcul de l'indice de masse musculaire squelettique. Le pipeline présenté a ensuite été utilisé pour évaluer la sarcopénie pour le pronostic des patients atteints de cancers solides en état critique, validant ainsi la viabilité pratique de notre approche DRL dans des contextes cliniques réels.La deuxième partie de cette thèse se concentre sur le renforcement de la sécurité et de la fiabilité des modèles d'IA en incorporant une robustesse certifiée contre les attaques adversariales. Les attaques adversariales ont mis en évidence la vulnérabilité des modèles d'apprentissage profond, en particulier dans des domaines à haut risque comme le diagnostic médical. Pour atténuer cette vulnérabilité, nous explorons et développons des approches garantissant un niveau de robustesse contre les manipulations adversariales en tirant parti de la puissance des modèles de diffusion et du lissage aléatoire. Ces méthodes fournissent non seulement une protection contre les éventuelles violations de sécurité, mais instaurent aussi une plus grande confiance dans le déploiement des systèmes de santé pilotés par l'IA.À travers ces orientations de recherche, cette thèse contribue à l'évolution continue de l'apprentissage profond en médecine en abordant les défis de la rareté des données et des attaques adversariales, traçant ainsi une voie potentielle vers un avenir plus résilient et plus fiable pour l'IA dans l'imagerie médicale.", "source_url": "https:\/\/openalex.org\/W7152083672", "pdf_url": "", "doi": "10.70675\/4ffc4285z667dz430czb4d9za731fb5f4d27", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7152083672", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 88, "title": "Semi-Supervised Deep Learning using Consistency-Based Methods for Segmentation of Medical Images", "title_ar": "", "title_en": "Semi-Supervised Deep Learning using Consistency-Based Methods for Segmentation of Medical Images", "author": "Axel Rönnberg", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "KTH Royal Institute of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; ai-applications; self-semi-supervised; predictive-analytics; image-segmentation; medical-imaging; ai-education", "keywords": "", "abstract": "In radiation therapy, a form of cancer treatment, accurately locating the anatomical structures is required in order to limit the impact on healthy cells. The automatic task of delineating these structures and organs is called segmentation, where each pixel in an image is classified and assigned a label. Recently, deep neural networks have proven to be efficient at automatic medical segmentation. However, deep learning requires large amounts of training data. This is a restricting feature, especially in the medical field due to factors such as patient confidentiality. Nonetheless, the main challenge is not the image data itself but the lack of high-quality annotations. It is thus interesting to investigate methods for semi-supervised learning, where only a subset of the images re- quires annotations. This raises the question if these methods can be acceptable for organ segmentation, and if they will result in an increased performance in comparison to supervised models. A category of semi-supervised methods applies the strategy of encouraging consistency between predictions. Consistency Training and Mean Teacher are two methods in which the network weights are updated in order to minimize the impact of input perturbations such as data augmentations. In addition, the Mean Teacher method trains two models, a Teacher and a Student. The Teacher is updated as an average of consecutive Student models, using Temporal Ensembling. To resolve the question whether semi-supervised learning could be beneficial, the two mentioned techniques are investigated. They are used in training deep neural networks with an U-net architecture to segment the bladder and anorectum in 3D CT images. The results showed signs of promise for Consistency Training and Mean Teacher, with nearly all model configurations having improved segmentations. Results also showed that the methods caused a reduction in performance variance, primarily by limiting poor delineations. With these results in hand, the use of semi-supervised learning should definitely be considered. However, since the segmentation improvement was not repeated in all experiment configurations, more research needs to be done.", "source_url": "https:\/\/openalex.org\/W3111652972", "pdf_url": "", "doi": "", "handle": "", "repository_name": "KTH Publication Database DiVA (KTH Royal Institute of Technology)", "source_record_id": "W3111652972", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 91, "title": "New and Advanced Explainable Artificial Intelligence Approaches for Managing Project Risks", "title_ar": "", "title_en": "New and Advanced Explainable Artificial Intelligence Approaches for Managing Project Risks", "author": "Bodrunnessa Badhon", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "generative-ai", "research_subdomain": "xai", "categories": "machine-learning; generative-ai; responsible-ai; expert-systems; ai-applications; predictive-analytics; xai; fairness; ai-smart-cities", "keywords": "", "abstract": "Project Risk Management (PRM) is a critical component of modern project governance, ensuring that uncertainties are systematically identified, assessed, and mitigated to minimize their impact on project outcomes. However, existing PRM practices remain largely subjective and qualitative, relying on expert judgment and static checklists that fail to capture dynamic interdependencies and evolving risk interactions. Recent advances in Artificial Intelligence (AI), particularly Machine Learning (ML) have improved predictive accuracy in PRM, but the opacity of many high-performing models has hindered adoption in safety-critical and managerial contexts where transparency, traceability, and stakeholder trust are essential. This thesis addresses these challenges by developing a suite of Explainable Artificial Intelligence (XAI) frameworks to support trustworthy project risk analysis. XAI has emerged as a promising approach to enhance transparency and interpretability in AI-based risk analysis; however, existing methods still exhibit several critical limitations that motivate this research. Current XAI techniques struggle to capture interrelations and cascading effects among risk factors, often producing fragmented explanations that overlook the structural dependencies underlying overall project vulnerability. In PRM, this limitation is particularly critical as risks are inherently interdependent, and disturbances in one aspect can propagate across multiple dimensions of project performance. Without incorporating such relationships into XAI models, the underlying causes of risk cannot be effectively identified, leading to explanations that lack depth and meaningful insight. Moreover, many post-hoc XAI models exhibit inconsistent or unstable explanations due to random neighborhood sampling or sensitivity to input variation, leading to unreliable outputs that are difficult to reproduce. In addition, existing XAI approaches remain largely static, offering one-way descriptions that overlook the temporal evolution and causal propagation of risks. Finally, they lack mechanisms for interactive engagement, preventing decision-makers from dynamically querying model behavior or exploring alternative “what-if” scenarios through conversational interfaces. Collectively, these limitations underscore the need for an integrated XAI-driven framework that unifies interpretability, scalability, causality, and interactivity. To address these shortcomings, this thesis defines four interrelated research objectives. Each objective focuses on a specific methodological limitation, namely interpretability, scalability, causality, and interactivity, while collectively forming a cohesive, progressive framework for explainable, trustworthy project risk management. The research commences by addressing the lack of interpretability and structured reasoning in existing risk models through the Interpretable Risk Assessment Framework based on Belief Rule-Based systems (IRAF-BRB). Traditional AI models for risk prediction often sacrifice interpretability for performance, making their outputs difficult for decision-makers to trust. IRAF-BRB resolves this issue by integrating expert knowledge with data-driven learning under explicit interpretability constraints. It ensures that model parameters remain meaningful, consistent with domain expertise, and explainable in linguistic terms. By optimizing belief rules using a modified Differential Evolution strategy while preserving human-interpretable reasoning, IRAF-BRB achieves both transparency and predictive reliability. This framework establishes the foundation for interpretable AI in PRM by showing that high accuracy can coexist with auditable, expert-aligned rule-based reasoning. While the IRAF-BRB ensures interpretability, it remains limited in scalability and adaptability to high-dimensional data. To overcome these limitations, the second framework, the Multi-Module Explainable Artificial Intelligence for Project Risk Management (MMXAI-PRM), extends explainability to complex, data-rich environments. MMXAI-PRM integrates generative modeling and post-hoc explanation to handle nonlinear relationships and class imbalance that traditional interpretable models cannot address. By combining synthetic data generation and local explanation, it enables consistent, stable interpretive reasoning even when training data are limited or heterogeneous. The framework provides richer explanatory insight by identifying key risk interactions and quantifying their relative influence, thereby enabling more informed prioritization and mitigation strategies. This contribution advances scalability and generalizability in explainable risk modeling, bridging the gap between interpretable logic-based systems and high-capacity learning models. Despite these improvements, earlier frameworks remain limited to static state-based representations that do not explicitly model the temporal evolution or causal propagation of risks.", "source_url": "https:\/\/openalex.org\/W7153988587", "pdf_url": "", "doi": "10.26190\/unsworks\/32194", "handle": "", "repository_name": "UNSWorks (UNSW Sydney)", "source_record_id": "W7153988587", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 92, "title": "Explainable Artificial Intelligence in the Image Domain and its Applications to the Medical Field", "title_ar": "", "title_en": "Explainable Artificial Intelligence in the Image Domain and its Applications to the Medical Field", "author": "Mirtha Lucas", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "DePaul University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; generative-ai; medical-ai; responsible-ai; self-semi-supervised; predictive-analytics; cnn; gan; medical-imaging; xai; fairness", "keywords": "", "abstract": "This dissertation investigates the development of Explainable Artificial Intelligence (XAI) methods for deep learning models in the image domain, with a particular focus on medical imaging applications. Although neural networks achieve high predictive performance, their lack of interpretability limits their adoption in critical domains such as healthcare, where transparency and trust are essential. This work addresses this challenge by proposing novel approaches that improve the interpretability and reliability of model predictions. A primary contribution is the introduction of Riemann–Stieltjes Integrated Grad-CAM (RSI Grad-CAM), a gradient-based attribution method that generates more relevant and spatially localized saliency maps. The method is evaluated using quantitative metrics and statistical analysis, demonstrating improved performance over established techniques such as Grad-CAM and Integrated Grad-CAM. A secondary contribution is the analysis of the interpretability of intermediate layers in convolutional neural networks, demonstrating that clinically meaningful features, such as spiculation in lung nodules, can be identified within specific layers of the model. A third contribution is the development of a human-centered framework based on a Variational Autoencoder-Auxiliary Classifier Wasserstein Generative Adversarial Network (VAE-AC-WGAN), which generates contrastive explanations aligned with clinical reasoning. Experimental results demonstrate that the proposed method produces higher-quality reconstructions and more informative anomaly maps than baseline approaches (e.g., medXGAN), as reflected in improved structural similarity (SSIM) and classifier-based evaluation metrics (AUROC). Furthermore, human evaluation indicates that the generated explanations better capture clinically relevant features, such as spiculation, supporting their interpretability in medical imaging contexts. Overall, this work advances XAI by enhancing model transparency and supporting the trustworthy integration of AI systems into medical decision-making.", "source_url": "https:\/\/openalex.org\/W7172068913", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Digital Commons-DePaul (DePaul University)", "source_record_id": "W7172068913", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 97, "title": "Explainable artificial intelligence for analysing mosquito flight behaviour", "title_ar": "", "title_en": "Explainable artificial intelligence for analysing mosquito flight behaviour", "author": "Yasser M. Qureshi", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Warwick", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "image-segmentation", "categories": "machine-learning; computer-vision; medical-ai; responsible-ai; predictive-analytics; optimization; image-segmentation; xai", "keywords": "", "abstract": "Mosquitoes are among the deadliest organisms on Earth, transmitting diseases that threaten global health. Malaria alone kills over half a million people annually, mostly young children. While insecticide-treated bed nets and indoor residual spraying have reduced transmission, these measures are increasingly compromised by insecticide-resistant mosquito populations and their behavioural adaptations. Despite well-documented genetic resistance, behavioural resistance remains poorly understood but critical to intervention efficacy. This thesis helps to address this gap by developing a machine learning and explainable AI (XAI) framework to analyse mosquito flight behaviours, focusing on identifying traits linked to resistance. The framework integrates preprocessing, trajectory segmentation, feature extraction, classification, and SHapley Additive exPlanations (SHAP) to provide interpretable insights. New techniques addressed tracking challenges like occlusions and spatial distortions, ensuring robustness. Three experimental conditions were analysed: (1) mating swarm data to classify males from non-males, (2) behavioural differences between insecticide-susceptible (IS) and insecticide-resistant (IR) mosquitoes near untreated nets, and (3) analogous trials involving permethrin-treated nets. The primary contributions of this thesis include the identification of behavioural signatures associated with insecticide resistance. IR mosquitoes exhibited reduced flight speeds, more direct paths, and lower fractal dimensions than IS mosquitoes, suggesting goal-directed flight. Exposure to permethrin-treated nets triggered erratic flight patterns in both IR and IS mosquitoes, with SHAP analysis identifying key features predictive of irritancy. Comparative analyses using this framework evaluated the performance of different tracking systems, confirming that 2D telecentric imaging matches 3D stereoscopic tracking performance, with singlecamera setups offering similar approximations at long ranges. Additionally, synthetic trajectories enhanced model interpretability by probing decision boundaries. The framework’s versatility and interpretability provide insights that could inform the design of interventions targeting mosquito vulnerabilities, with potential applications to other complex behavioural systems where understanding subtle patterns is essential.", "source_url": "https:\/\/openalex.org\/W7128281259", "pdf_url": "", "doi": "10.82444\/warw.33910837", "handle": "", "repository_name": "Warwick Research Archive Portal (University of Warwick)", "source_record_id": "W7128281259", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 101, "title": "Explainable artificial intelligence for integrity assessment of civil engineering infrastructure: Insights from various engineering applications", "title_ar": "", "title_en": "Explainable artificial intelligence for integrity assessment of civil engineering infrastructure: Insights from various engineering applications", "author": "Masoud Haghbin", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Granada", "university_source": "repository_host", "college": "", "department": "", "country": "Spain", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "computer-vision", "research_subdomain": "xai", "categories": "machine-learning; computer-vision; responsible-ai; predictive-analytics; xai; fairness", "keywords": "", "abstract": "The incorporation of Artificial Intelligence (AI) into civil engineering, especially structural engineering, has the potential to improve decision-making by making it more precise and data driven. However, relying on AI to make decisions without understanding its reasoning poses a serious risk: It undermines the essential concept of human judgment. This concern has led to global regulatory measures, including the European Union AI Act, which highlights the importance of explainability and human oversight in creating trustworthy AI. In response to these needs, eXplainable AI (XAI) has attracted significant attention from researchers over the past five years. It serves as a technical tool to shed light on the internal computing processes of AI systems, which are often seen as black boxes, addressing safety and ethical implications. Most research into the application of XAI in civil engineering is conducted under controlled and standardized conditions. However, real-world engineering problems often occur in complex and changing environments. These problems often span multiple domains, making their modeling challenging despite their importance. In addition, the presence of scarce and poor quality data and real-world behaviors that are difficult to replicate in laboratory experiments (e.g., in the presence of damage) add to these complications. These inherent issues challenge the fundamental assumptions of many AI models, which are usually trained and validated in controlled or synthetic environments. Consequently, both the predictive accuracy and the reliability of model explanations can deteriorate when these models are extended to real-world situations, a risk that is often underestimated or overlooked. In 2026, the European Union’s AI Act will come into force for critical infrastructures, such as the vast majority of civil engineering infrastructure, making explainability and human oversight a legal requirement. There is a significant technical gap, as well as pressing safety, ethical, and regulatory concerns, due to the lack of assessments of the robustness of XAI models in real-world applications. This thesis shifts the focus from presenting novel AI models that excel in standardized benchmark problems to understanding how cutting-edge but existing AI and XAI methods perform under the irregular and data-poor conditions typically found in real-world civil engineering systems. This research highlights the need for realistic domain-aware deployment strategies that prioritize robustness, interpretability, and practical insights over purely theoretical advances. Rather than relying on idealized datasets or conditions favorable to AI, such as clean, balanced, and well-annotated data, which do not accurately represent realworld engineering conditions, this thesis focuses on investigating, understanding, and improving the behavior of the AI model in complex environments using real data from various case studies. To achieve these objectives, the thesis begins by investigating a complex type of rail surface degradation problem known as corrugation. This phenomenon is affected by a combination of dynamic and static factors, including vehicle-track interaction, material properties, and environmental conditions. However, measuring or integrating these contributing variables presents significant challenges as they are often difficult to assess directly, expensive, or even unfeasible to acquire in real-world situations. In order to ensure a realistic and application-oriented approach, a minimal set of input parameters (acceleration and velocity), which are readily available and can be easily measured from the moving vehicle, have been selected to predict the rail corrugation profile based on time series data of these input parameters. To align with practical constraints in the field, multiple damaged zones were considered in order to capture the variability and complexity of actual conditions of the rail infrastructure. Despite the common practice of removing outliers as an initial step in data preprocessing, the research presented in this thesis intentionally refrains from doing so in this particular scenario. Retaining outliers is crucial for accurately representing the unpredictable nature of real-world engineering data, where noise, anomalies, and irregular patterns are inherent in reality. Excluding them could lead to an overly simplified understanding of the problem and diminish the relevance of AI models in practical applications. The XAI method, Grad-CAM, is applied to enhance transparency and increase human oversight in the prediction of rail corrugation under the aforementioned realistic conditions. This approach allows for visual interpretation of the model decision-making process, helping to bridge the gap between black-box predictions and domain-relevant understanding.", "source_url": "https:\/\/openalex.org\/W7154665432", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Institutional Repository of the University of Granada (University of Granada)", "source_record_id": "W7154665432", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 102, "title": "Bridging the Gap: From Artificial Intelligence and Optimization Theory to Action", "title_ar": "", "title_en": "Bridging the Gap: From Artificial Intelligence and Optimization Theory to Action", "author": "Periklis Petridis", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Massachusetts Institute of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "xai", "categories": "machine-learning; generative-ai; medical-ai; responsible-ai; ai-applications; predictive-analytics; xai; ai-industry; ai-smart-cities", "keywords": "", "abstract": "Despite significant theoretical advances in Operations Research (OR) and Artificial Intelligence (AI), a persistent gap remains between these developments and their practical implementation in real-world settings. Despite significant progress in these fields, many OR and ML approaches struggle to scale to realistic problem sizes, lack robustness to uncertainty, or fail to address implementation constraints faced by practitioners in industry. Through four distinct works conducted in collaboration with industry partners, this research demonstrates how methodological advancements can bridge this theory-practice divide while maintaining rigorous theoretical foundations and guarantees. In the first part, we focus on optimization methodologies that scale traditional OR approaches to handle real-world problem sizes and uncertainty. In Chapter 2, we develop a stochastic Benders decomposition scheme for large-scale network design problems, a class of problems ubiquitous in logistics, transportation, and energy sectors. By incorporating sampling techniques within the decomposition framework, we achieve deterministic optimality guarantees while reducing computational costs, enabling solutions for networks with 700 nodes—an order of magnitude larger than previously tractable instances—while achieving optimality gaps of 5-7% compared to 16-27% for traditional deterministic approaches. In Chapter 3, we present a holistic framework for industrial decarbonization, developed with a major phosphate producer planning to quadruple energy consumption while transitioning to renewable sources. Our robust optimization approach combines strategic capacity expansion planning over a 25-year horizon with adaptive operational models, providing 95% reliability guarantees while balancing solar and wind integration with battery storage to meet a projected 12 TWh annual demand. In the second part, we shift our focus to developing AI systems that address the unique challenges of medical data abstraction and clinical decision support. In Chapter 4, we address the challenge of automating clinical data abstraction from electronic health records, collaborating with the Society of Thoracic Surgeons to populate their Adult Cardiac Surgery Database. Our AI pipeline combines 31 models per target variable with a two-tiered quality control system, achieving over 99% accuracy while automatically extracting 43-50% of registry variables, demonstrating how AI can dramatically reduce manual abstraction burden while maintaining clinical standards. In Chapter 5, we extend this healthcare AI focus by developing xHAIM (Explainable Holistic AI in Medicine), which addresses the limitations of current clinical AI systems in handling extensive patient records, providing interpretability, and incorporating medical knowledge. Through semantic similarity techniques and generative AI, xHAIM improves predictive performance while generating clinically grounded explanations that enhance trust and adoption by healthcare practitioners.", "source_url": "https:\/\/openalex.org\/W7123375454", "pdf_url": "", "doi": "", "handle": "", "repository_name": "DSpace@MIT (Massachusetts Institute of Technology)", "source_record_id": "W7123375454", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 105, "title": "GENERATIVE LARGE LANGUAGE MODELS FOR TRANSPARENT ARTIFICIAL INTELLIGENCE IN CLINICAL RESEARCH: ENHANCING INTERPRETABILITY THROUGH APPRAISAL AND EXPLANATION", "title_ar": "", "title_en": "GENERATIVE LARGE LANGUAGE MODELS FOR TRANSPARENT ARTIFICIAL INTELLIGENCE IN CLINICAL RESEARCH: ENHANCING INTERPRETABILITY THROUGH APPRAISAL AND EXPLANATION", "author": "Fangwen Zhou", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "McMaster University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; nlp; generative-ai; medical-ai; responsible-ai; transfer-learning; predictive-analytics; transformers; llm; xai", "keywords": "", "abstract": "Background The rapid growth of medical literature necessitates effective, transparent automation tools for classification. Generative large language models (LLMs), including the Generative Pre-trained Transformer (GPT), have the potential to provide transparent classification and explain other black box models. Objective This sandwich thesis evaluates the performance of GPT in 1) classifying biomedical literature compared with a fine-tuned BioLinkBERT model, and 2) explaining the decision of encoder-only models with feature attributions compared to traditional eXplainable AI (XAI) frameworks like SHapley Additive exPlanations (SHAP) and integrated gradients (IG). Methods Randomly sampled, manually annotated clinical research articles from the Health Information Research Unit (HIRU) were used along with a top-performing BioLinkBERT classifier. In Chapter 2, GPT-4o and GPT-o3-mini were used either alone or with BioLinkBERT’s predictions in the prompt to classify article methodological rigour based on HIRU’s criteria. Either the title and abstract or the full text was provided to GPT. Performance was compared to the BioLinkBERT model and assessed primarily using Matthew’s correlation coefficient (MCC). In Chapter 3, GPT-4o was used to generate feature attributions for the BioLinkBERT model through masking perturbations and was compared to SHAP and IG using a modified area under the perturbation curve (AOPC) metric which gives a measure of performance. Results GPT-4o alone, using full text (MCC 0.429), achieved comparable classification performance to BioLinkBERT (MCC 0.466). Performance was worse with other models and inputs. As a perturbation explainer, GPT-4o’s (AOPC 0.029) performance was poor and significantly underperformed compared to SHAP (AOPC 0.222) and IG (AOPC 0.225). The identified important tokens by GPT did not align with the manual appraisal criteria. Conclusion GPT has potential in appraising biomedical literature, even without explicit training. GPT’s transparency through textual explanations improves interpretability. GPT’s poor performance in generating faithful feature attributions warrants future research. The inherent variability and stochasticity of GPT outputs necessitate careful prompting and reproducibility measures.", "source_url": "https:\/\/openalex.org\/W7214197727", "pdf_url": "", "doi": "10.71548\/13232", "handle": "", "repository_name": "MacSphere (McMaster University)", "source_record_id": "W7214197727", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 107, "title": "An EXplainable Artificial Intelligence Credit Rating System", "title_ar": "", "title_en": "", "author": "Ayoub El-Qadi", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2023, "language": "fr", "research_domain": "nlp", "research_subdomain": "xai", "categories": "machine-learning; nlp; responsible-ai; xai", "keywords": "", "abstract": "Explicabilité d'un système de notation de risque de crédit basé sur l'intelligence artificielle Au cours des dernières années, le déficit de financement du commerce a atteint le chiffre alarmant de 1 500 milliards de dollars, soulignant une crise croissante dans le commerce mondial. Ce déficit est particulièrement préjudiciable aux petites et moyennes entreprises (PME), qui éprouvent souvent des difficultés à accéder au financement du commerce. Les systèmes traditionnels d'évaluation du crédit, qui constituent l'épine dorsale du finance-ment du commerce, ne sont pas toujours adaptés pour évaluer correctement la solvabilité des PME. Le terme \"credit scoring\" désigne les méthodes et techniques utilisées pour évaluer la solvabilité des individus ou des entreprises. Le score généré est ensuite utilisé par les institutions financières pour prendre des décisions sur l'approbation des prêts, les taux d'intérêt et les limites de crédit. L'évaluation du crédit présente plusieurs caractéristiques qui en font une tâche difficile. Tout d'abord, le manque d'explicabilité des modèles complexes d'apprentissage automatique entraîne souvent une moindre acceptation des évaluations de crédit, en particulier parmi les parties prenantes qui exigent un processus décisionnel transparent. Cette opacité peut constituer un obstacle à l'adoption généralisée de techniques d'évaluation avancées. Un autre défi important est la variabilité de la disponibilité des données entre les pays et les dossiers financiers souvent incomplets des PME, ce qui rend difficile le développement de modèles universellement applicables. Dans cette thèse, nous avons d'abord abordé la question de l'explicabilité en utilisant des techniques de pointe dans le domaine de l'intelligence artificielle explicable (XAI). Nous avons introduit une nouvelle stratégie consistant à comparer les explications générées par les modèles d'apprentissage automatique avec les critères utilisés par les experts en crédit. Cette analyse comparative a révélé une divergence entre le raisonnement du modèle et le jugement de l'expert, soulignant la nécessité d'incorporer les critères de l'expert dans la phase de formation du modèle. Les résultats suggèrent que l'alignement des explications générées par la machine sur l'expertise humaine pourrait être une étape cruciale dans l'amélioration de l'acceptation et de la fiabilité du modèle.Par la suite, nous nous sommes concentrés sur le défi que représentent les don-nées financières éparses ou incomplètes. Nous avons incorporé des évaluations de crédit textuelles dans le modèle d'évaluation du crédit en utilisant des techniques de pointe de traitement du langage naturel (NLP). Nos résultats ont démontré que les modèles formés à la fois avec des données financières et des évaluations de crédit textuelles étaient plus performants que ceux qui s'appuyaient uniquement sur des données financières. En outre, nous avons montré que notre approche pouvait effectivement générer des scores de crédit en utilisant uniquement des évaluations de risque textuelles, offrant ainsi une solution viable pour les scénarios dans lesquels les mesures financières traditionnelles ne sont pas disponibles ou insuffisantes", "source_url": "https:\/\/openalex.org\/W7151827639", "pdf_url": "https:\/\/theses.hal.science\/tel-04472510v1\/document", "doi": "10.70675\/2d09695dz9d70z4602zb460zfd801584e220", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7151827639", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:46", "updated_at": "2026-11-15 16:52:46" }, { "id": 111, "title": "Explainable artificial intelligence (XAI) for healthcare decision-making", "title_ar": "", "title_en": "Explainable artificial intelligence (XAI) for healthcare decision-making", "author": "Zhiwei Zeng", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Nanyang Technological University", "university_source": "repository_host", "college": "", "department": "", "country": "Singapore", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "medical-ai", "research_subdomain": "xai", "categories": "machine-learning; medical-ai; responsible-ai; ai-applications; self-semi-supervised; xai; fairness; ai-cybersecurity", "keywords": "", "abstract": "With the increasing ubiquity of Artificial Intelligence (AI) in daily life and into more critical applications, explainability has been brought under the spotlight due to ethical, social, and legal concerns. Explainability is particularly important in application domains involving high-stakes decision-making, such as healthcare, medicine, law, and defence, as simply having high performance may not satisfy all the desiderata of stakeholders. Stakeholders in these domains often require greater insights into the decision-making process, better support for tracing decision trails and identifying errors, as well as greater trustworthiness from an AI system. An eXplainable AI (XAI) is one that has the ability to explain the outputs or functioning of a model in a clear and understandable manner to a human. In the explanation process of XAI, explainability approaches provide explanatory information to human stakeholders in order to facilitate their understanding of the outputs and functioning of an AI model. The renewed understanding then affects the extent to which the desiderata are satisfied. Over the past decade, a plethora of explainability approaches have been proposed, ranging from ante-hoc methods that seek to build explainability features into a model itself to post-hoc methods that use techniques to improve the explainability of an already developed model. Despite the rapid growth of the XAI field, three key open problems emerge from a review of the existing XAI methods. Firstly, post-hoc methods have been increasingly used for high-stakes decisions, but most of them provide explanatory information that is not faithful or not insightful to what the original model computes. Secondly, most existing methods can only provide associational explanations and lack the structure and reasoning ability to provide contrastive and counterfactual explanations. Thirdly, most existing methods do not provide explanations to suit different levels of understanding. Research problem 1 - Post-hoc explainability methods have quickly gained popularity as they are mostly model-agnostic and highly extensible. However, without elucidating or understanding the model’s inner workings, explanations generated by post-hoc methods may not faithfully represent the underlying relationships of the model and often do not provide enough information to facilitate the understanding of how the output is obtained. In this thesis, I propose an ante-hoc explainability methodology to support the making and explaining of high-stakes decisions in healthcare. It is an argumentative methodology that can infuse explainability into the model and generate explanations that are faithful and insightful to the inner workings and internal representations of the model. It follows a “debate-like” process that shares many similarities with the way humans deliberate and make decisions, thereby giving it unique advantages in generating human-understandable explanations. I instantiate the argumentative methodology in three explainable decision-making approaches for the modelling and reasoning of contexts, qualitative preferences as well as both attack and support relationships. Research problem 2 - Although contrastive and counterfactual explanations have attracted more research interest recently, existing XAI methods mostly generate associational explana- tions. Most existing methods are good at answering “Why X? ” but poor at answering questions that require higher-level inference, such as “Why not-X? ”. This can be attributed to the fact that most of them lack the necessary structures and mechanisms to perform high-level logical reasoning. Building on the structures and reasoning ability of the three proposed explain- able decision-making approaches, I construct different forms of contrastive and counterfactual explanations that can answer questions such as “Why X instead of Y? ” and “Why not-X? ”. Research problem 3 - Different levels of understanding are often required in most practical scenarios. However, most existing XAI methods do not provide explanations that can cater to the different levels of understanding. In this thesis, I propose and study formal properties of explanatory information in order to generate such explanations. I propose notions and computational constructs for generating explanations that can provide different amounts of justification or have different dialectical strengths, as well as focused explanations that only include selective information in the decision-making process. As a case study, I implement a proposed explainable decision-making approach for the diagnostics and prognostics of Alzheimer’s Disease (AD). Empirical evaluation with real-world AD datasets and human evaluations were conducted to assess the performance and explainability of the approach.", "source_url": "https:\/\/openalex.org\/W4224077403", "pdf_url": "https:\/\/dr.ntu.edu.sg\/bitstream\/10356\/155849\/4\/Thesis_final_Zhiwei%20Zeng_signed%20with%20watermark.pdf", "doi": "10.32657\/10356\/155849", "handle": "", "repository_name": "DR-NTU (Nanyang Technological University)", "source_record_id": "W4224077403", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 117, "title": "Interpretable Measures for Arabic Readability", "title_ar": "", "title_en": "Interpretable Measures for Arabic Readability", "author": "Rabih, Nour", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United Arab Emirates", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; nlp; responsible-ai; expert-systems; ai-applications; predictive-analytics; transformers; arabic-nlp; xai; ai-education", "keywords": "", "abstract": "Readability assessment is a crucial task in natural language processing (NLP), with wide reaching implications in education, accessibility, and language learning. While prior research has primarily focused on predicting readability levels, this work shifts the focus toward understanding why a text is assigned a particular level. By identifying key linguistic features that drive readability, the goal is to enhance interpretability in Arabic text complexity assessment. Arabic poses unique challenges for automatic readability evaluation due to its morphological richness, variation across dialects, and limited availability of annotated corpora. The complex structure of Arabic words, and extensive derivational morphology add layers of difficulty in determining textual complexity. These challenges necessitate a more comprehensive approach to readability evaluation beyond traditional classification. To address these limitations, this work leverages the Balanced Arabic Readability Evaluation Corpus (BAREC)—a large-scale, manually annotated dataset of 68,182 sentences across 19 readability levels, from kindergarten to postgraduate comprehension. BAREC is designed to ensure genre diversity, topical variety, and balanced target audiences, making it a strong foundation for Arabic readability modeling. This study systematically engineers and evaluates a comprehensive set of linguistic features from BAREC’s annotation guidelines, including morphology, syntax, vocabulary, syllables, and content and vocabulary-based measures. Through extensive experimentation, it investigates the effectiveness of these features in explaining and predicting readability levels across three task configurations: foundational levels (1–11), advanced levels (12–19), and the full 19-level scale. In this study, we explore a variety of modeling approaches, including traditional machine learning classifiers, a fully interpretable rule-based model, and an AraBERT model based on transformers. Furthermore, a hybrid model is proposed that integrates linguistic features with contextual embeddings to improve both performance and transparency. The results demonstrate that linguistic features significantly enhance interpretability, especially in low-resource settings, and that feature-driven models can provide human readable justifications aligned with expert annotation practices. This research not only advances Arabic readability prediction but also contributes interpretable insights into text complexity, paving the way for more accessible and pedagogically meaningful NLP tools in Arabic education.", "source_url": "https:\/\/openalex.org\/W7106299741", "pdf_url": "", "doi": "10.82419\/276", "handle": "", "repository_name": "MBZUAI iRep", "source_record_id": "W7106299741", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 119, "title": "Arabic dialect texts classification", "title_ar": "", "title_en": "Arabic dialect texts classification", "author": "Areej O. Alshutayri", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "nlp", "research_subdomain": "arabic-nlp", "categories": "machine-learning; nlp; arabic-nlp", "keywords": "", "abstract": "This study investigates how to classify Arabic dialects in text by extracting features which show the differences between dialects. There has been a lack of research about classification of Arabic dialect texts, in comparison to English and some other languages, due to the lack of Arabic dialect text corpora in comparison with what is available for dialects of English and some other languages. What is more, there is an increasing use of Arabic dialects in social media, so this text is now considered quite appropriate as a medium of communication and as a source of a corpus. We collected tweets from Twitter, comments from Facebook and online newspapers from five groups of Arabic dialects: Gulf, Iraqi, Egyptian, Levantine, and North African. The research sought to: 1) create a dataset of Arabic dialect texts to use in training and testing the system of classification, 2) find appropriate features to classify Arabic dialects: lexical (word and multi-word-unit) and grammatical variation across dialects, 3) build a more sophisticated filter to extract features from Arabic-character written dialect text files. In this thesis, the first part describes the research motivation to show the reason for choosing the Arabic dialects as a research topic. The second part presents some background information about the Arabic language and its dialects, and the literature review shows previous research about this subject. The research methodology part shows the initial experiment to classify Arabic dialects. The results of this experiment showed the need to create an Arabic dialect text corpus, by exploring Twitter and online newspaper. The corpus used to train the ensemble classifier and to improve the accuracy of classification the corpus was extended by collecting tweets from Twitter based on the spatial coordinate points and comments from Facebook posts. The corpus was annotated with dialect labels and used in automatic dialect classification experiments. The last part of this thesis presents the results of classification, conclusions and future work.", "source_url": "https:\/\/openalex.org\/W2947159833", "pdf_url": "http:\/\/etheses.whiterose.ac.uk\/23600\/1\/Areej%20Alshutayri%20Thesis.pdf", "doi": "", "handle": "", "repository_name": "White Rose eTheses Online (University of Leeds, The University of Sheffield, University of York)", "source_record_id": "W2947159833", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 120, "title": "Enhanced root extraction and document classification algorithm for Arabic text", "title_ar": "", "title_en": "Enhanced root extraction and document classification algorithm for Arabic text", "author": "Amal Alsaad", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Brunel University London", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2016, "language": "en", "research_domain": "nlp", "research_subdomain": "arabic-nlp", "categories": "machine-learning; deep-learning; nlp; arabic-nlp", "keywords": "", "abstract": "Many text extraction and classification systems have been developed for English and other international languages; most of the languages are based on Roman letters.However, Arabic language is one of the difficult languages which have special rules and morphology.Not many systems have been developed for Arabic text categorization.Arabic language is one of the Semitic languages with morphology that is more complicated than English.Due to its complex morphology, there is a need for pre-processing routines to extract the roots of the words then classify them according to the group of acts or meaning.In this thesis, a system has been developed and tested for text classification.The system is based on two stages, the first is to extract the roots from text and the second is to classify the text according to predefined categories.The linguistic root extraction stage is composed of two main phases.The first phase is to handle removal of affixes including prefixes, suffixes and infixes.Prefixes and suffixes are removed depending on the length of the word, while checking its morphological pattern after each deduction to remove infixes.In the second phase, the root extraction algorithm is formulated to handle weak, defined, eliminated-long-vowel and two-letter geminated words, as there is a substantial great amount of irregular Arabic words in texts.Once the roots are extracted, they are checked against a predefined list of 3800 triliteral and 900 quad literal roots.Series of experiments has been conducted to improve and test the performance of the proposed algorithm.The obtained results revealed that the developed algorithm has better accuracy than the existing stemming algorithm.The second stage is the document classification stage.In this stage two nonparametric classifiers are tested, namely Artificial Neural Networks (ANN) and Support Vector Machine (SVM).The system is trained on 6 categories: culture, economy, international, local, religion and sports.The system is trained on 80% of the available data.From each category, the 10 top frequent terms are selected as features.Testing the classification algorithms has been done on the remaining 20% of the documents.The results of ANN and SVM are compared to the standard method used for text classification, the terms frequency-based method.Results show that ANN and SVM have better accuracy (80-90%) compared to the standard method (60-70%).The proposed method proves the ability to categorize the Arabic text documents into the appropriate categories with a high precision rate.", "source_url": "https:\/\/openalex.org\/W2556582302", "pdf_url": "http:\/\/bura.brunel.ac.uk\/bitstream\/2438\/13510\/1\/FulltextThesis.pdf", "doi": "", "handle": "", "repository_name": "Brunel University Research Archive (BURA) (Brunel University London)", "source_record_id": "W2556582302", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 121, "title": "Sentiment analysis of dialectical Arabic social media content using a hybrid linguistic-machine learning approach", "title_ar": "", "title_en": "Sentiment analysis of dialectical Arabic social media content using a hybrid linguistic-machine learning approach", "author": "GN Alwakid", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "nlp", "research_subdomain": "sentiment-analysis", "categories": "machine-learning; nlp; sentiment-analysis; arabic-nlp", "keywords": "", "abstract": "Despite the enormous increase in the number of Arabic posts on social networks, the sentiment analysis research into extracting opinions from these posts lags behind that for the English language. This is largely attributed to the challenges in processing the morphologically complex Arabic natural language and the scarcity of Arabic NLP tools and resources. This complex task is further exacerbated when analysing dialectal Arabic that do not abide by the formal grammatical structure. Based on the semantic modelling of the target domain’s knowledge and multi-factor lexicon-based sentiment analysis, the intent of this research is to use a hybrid approach, integrating linguistic and machine learning methods for sentiment analysis classification of dialectal Arabic. First, a dataset of dialectal Arabic tweets was collected focusing on the unemployment domain, which is annotated manually. The tweets cover different dialectal Arabic in Saudi Arabia for which a comprehensive Arabic sentiment lexicon was constructed. This approach to sentiment analysis also integrated a novel light stemming mechanism towards improved Saudi dialectal Arabic stemming. Subsequently, a novel multi-factor lexicon-based sentiment analysis algorithm was developed for domain-specific social media posts written in dialectal Arabic. The algorithm considers several factors (emoji, intensifiers, negations, supplications) to improve the accuracy of the classifications. Applying this model to a central problem of sentiment analysis in dialectical Arabic, these operational techniques were deployed in order to assess analytical performance across social media channels which are vulnerable to semantic and colloquial variations. Finally, this study presented a new hybrid approach to sentiment analysis where domain knowledge is utilised in two methods to combine computational linguistics and machine learning; the first method integrates the problem domain semantic knowledgebase in the machine learning training features set, while the second uses the outcome of the lexicon-based sentiment classification in the training of the machine learning methods. By integrating these techniques into a single, hybridised solution, a greater degree of accuracy and consistency was achieved than applying each approach independently, confirming a pragmatic solution to sentiment classification in dialectical Arabic text.", "source_url": "https:\/\/openalex.org\/W3158386543", "pdf_url": "http:\/\/irep.ntu.ac.uk\/id\/eprint\/42474\/1\/Ghadah%20Alwakid%202021.pdf", "doi": "", "handle": "", "repository_name": "Nottingham Trent University's Institutional Repository (Nottingham Trent Repository)", "source_record_id": "W3158386543", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 124, "title": "Aspect-based Sentiment Analysis for Arabic Reviews Using Deep Learning", "title_ar": "", "title_en": "Aspect-based Sentiment Analysis for Arabic Reviews Using Deep Learning", "author": "Nada Almani", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "The University of Surrey", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "nlp", "research_subdomain": "sentiment-analysis", "categories": "machine-learning; deep-learning; computer-vision; nlp; responsible-ai; transfer-learning; sentiment-analysis; arabic-nlp; xai", "keywords": "", "abstract": "Sentiment analysis is a branch of machine learning that concerns about finding and classifying the polarity for given text. Because of the availability of huge amount of opinionated data that need to be analyzed and interpreted, a lot of recent machine learning research is focused on sentiment analysis applications. it gained a lot of interest due to the. Many sentiment analysis systems are modeled by using different machine learning techniques, but recently, deep learning, by using Artificial Neural Network (ANN) architecture, has showed significant improvements with high tendency to reveal the underlying semantic meaning in the input text. However, the output of these models could not be explained and the efficiency could not be analyzed because ANN models are considered as a black box and the success of these models comes at the cost of interpretability. The main objective of the presented work is developing Arabic sentiment analysis system that understands semantics in input reviews without using any linguistic resources. The first proposed model is Deep Attention-based Review Level Sentiment Analysis model (DARLSA) that use binary classifier to detect reviews’ polarities. Different scenarios and architectures were examined to test the ability of the proposed model to extract salient words out of the input. The results proved the ability of the proposed model to understand a given review by highlighting the most informative words to the class label. The model detected Arabic natural language linguistic features, such as intensification and negation styles, efficiently. Also, the effect of applying transfer learning technique on the problem of Arabic sentiment analysis is experimented on review level model. The second proposed architecture is Deep Attention-based Aspect Level Sentiment Analysis model (DAALSA) for classifying reviews polarity with respect to an aspect into three classes, positive, negative and neutral. Different models were proposed to test the effect of using different attention scoring functions on the classification performance. The results distinguished one model with superior performance compared to other proposed models. To obtain intuitive explanation of the trained models, both models are enhanced with visualization option. The final review representation is a distributed dense vector generated after passing through multi-layers neural network. Heatmap representation is used to visualize the final review representation. In addition, the attention layer’s scoring vector is visualized as well.", "source_url": "https:\/\/openalex.org\/W2996475580", "pdf_url": "", "doi": "10.15126\/thesis.00852989", "handle": "", "repository_name": "Surrey Research Insight Open Access (The University of Surrey)", "source_record_id": "W2996475580", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 128, "title": "Classification of medical documents and Opinions using machine learning", "title_ar": "", "title_en": "Classification of medical documents and Opinions using machine learning", "author": "Fatiha Youbi", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Algeria", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "nlp", "research_subdomain": "sentiment-analysis", "categories": "machine-learning; deep-learning; nlp; medical-ai; data-mining; sentiment-analysis", "keywords": "", "abstract": "In recent years, there has been increasing interest in the use of text mining (TM) and machine learning in healthcare. Document classi cation has been a common application, with many studies focusing on classifying medical reports from unstructured text data. However, there is also a need to utilize TM and machine learning for sentiment analysis of medical textual data in social networks and medical forums. In this thesis, the focus was on two TM applications in the medical domain: classifying autopsy reports to detect the manner of death in Wilaya of Tlemcen and analyzing patient and public opinions on healthcare and the COVID-19 pandemic using machine learning techniques. The experiments conducted in both studies showed that automated models for opinion analysis are task-speci c and that feature extraction and deep learning classi er architecture play important roles in the success of these models. The ndings could be useful for improving strategies related to drugs monitoring and COVID-19 surveillance. Future directions include exploring other types of deep learning techniques, using clinical documents for sentiment analysis, and analyzing Algerian health status based on machine learning and deep learning classi ers.", "source_url": "https:\/\/openalex.org\/W7135334549", "pdf_url": "https:\/\/dspace.univ-tlemcen.dz\/bitstreams\/c81c9cdb-90b6-4887-a5e6-fb6fcf10f7ad\/download", "doi": "", "handle": "", "repository_name": "Dépôt Institutionnel de lUniversité de Tlemcen", "source_record_id": "W7135334549", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 131, "title": "Ontology learning from the Arabic text of the Qur’an : concepts identification and hierarchical relationships extraction", "title_ar": "", "title_en": "Ontology learning from the Arabic text of the Qur’an : concepts identification and hierarchical relationships extraction", "author": "SM Alrehaili", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "nlp", "research_subdomain": "arabic-nlp", "categories": "machine-learning; nlp; ai-applications; arabic-nlp; ai-education", "keywords": "", "abstract": "Recent developments in ontology learning have highlighted the growing role ontologies play in linguistic and computational research areas such as language teaching and natural language processing. The ever-growing availability of annotations for the Qur’an text has made the acquisition of the ontological knowledge promising. However, the availability of resources and tools for Arabic ontology is not comparable with other languages. Manual ontology development is labour-intensive, time-consuming and it requires knowledge and skills of domain experts. This thesis aims to develop new methods for Ontology learning from the Arabic text of the Qur’an, including concepts identification and hierarchical relationships extraction. The thesis presents a methodology for reducing human intervention in building ontology from Classical Arabic Language of the Qur’an text. The set of concepts, which is a crucial step in ontology learning, was generated based on a set of patterns made of lexical and inflectional information. The concepts were identified based on adapted weighting schema that exploit a combination of knowledge to learn the relevance degree of a term. Statistical, domain-specific knowledge and internal information of Multi-Word Terms (MWTs) were combined to learn the relevance of generated terms. This methodology which represents the major contribution of the thesis was experimentally investigated using different terms generation methods. As a result, we provided the Arabic Qur’anic Terms (AQT) as a training resource for machine learning based term extraction. This thesis also introduces a new approach for hierarchical relations extraction from Arabic text of the Qur’an. A set of hierarchical relations occurring between identified concepts are extracted based on hybrid methods including head-modifier, set of markers for copula construct in Arabic text, referents. We also compared a number of ontology alignment methods for matching ontological bilingual Qur’anic resources. In addition, a multi-dimensional resource named Arabic Qur’anic Database (AQD) about the Qur’an is made for Arabic computational researchers, allowing regular expression query search over the included annotations. The search tool was successfully applied to find instances for a given complex rule made of different combined resources.", "source_url": "https:\/\/openalex.org\/W2806278576", "pdf_url": "http:\/\/etheses.whiterose.ac.uk\/20309\/1\/Alrehaili_SM_Computing_PhD2017.pdf", "doi": "", "handle": "", "repository_name": "White Rose eTheses Online (University of Leeds, The University of Sheffield, University of York)", "source_record_id": "W2806278576", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 132, "title": "Detection of anti-social behaviour in online communication in Arabic", "title_ar": "", "title_en": "Detection of anti-social behaviour in online communication in Arabic", "author": "Azalden Alakrot", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Limerick", "university_source": "repository_host", "college": "", "department": "", "country": "Ireland", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "nlp", "research_subdomain": "arabic-nlp", "categories": "machine-learning; nlp; medical-ai; ai-applications; arabic-nlp; ai-education", "keywords": "", "abstract": "Anti-social behaviour on social media cannot be easily ignored as it affects a large and growing percentage of the world’s population. It often has a nega tive effect on people’s lives; incidents of online abuse that may seem insignif icant can have a cumulative impact on mental health. An increasing num ber of incidents of suicide and violence have been reportedly provoked by anti-social behaviour on social media. Most of the existing machine-learning approaches for detection of offensive language are specifically tailored for online communication in English. Solutions targeting Arabic language are rare, while, as we also demonstrate in this thesis, offensive language is wide spread in Arabic social media as well. Our hypothesis has been that Arabic may require a specific approach different from the solutions for English due to the specific linguistic characteristics of Arabic text and the unique to Ara bic mixture of dialects frequently observed within the same conversation on social media. The objective of this thesis is to contribute to the work on the automatic pre vention of anti-social behaviour in online written communication in Ara bic by introducing a large dataset of YouTube comments and proposing a text-mining pipe-line for training a binary classifier. The main challenge to automatic detection of offensive language is the absence of appropriate training datasets. Thus, as part of this work we undertook to collect data from Arabic social media (Arabic YouTube channels) and construct a labelled dataset. Then we utilised this dataset to experiment with a variety of text pre processing techniques, feature-selection methods, and classification machine learning algorithms in order to recommend a process for automatic detection of offensive language in online written communication in Arabic. Our results are encouraging; they suggest Support Vector Machines classifier can be ef fectively deployed for the detection of offensive language in online written communication in Arabic. We believe that the proposed text-mining process will open the door for further research in this direction and will eventually result in effective automatic prevention of incidents of verbal abuse on Ara bic social media.", "source_url": "https:\/\/openalex.org\/W3139753181", "pdf_url": "", "doi": "10.34961\/3378", "handle": "", "repository_name": "University of Limerick Institutional Repository (University of Limerick)", "source_record_id": "W3139753181", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 135, "title": "Embedded Arabic text detection and recognition in videos", "title_ar": "", "title_en": "", "author": "Sonia Yousfi", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2016, "language": "fr", "research_domain": "nlp", "research_subdomain": "arabic-nlp", "categories": "deep-learning; computer-vision; nlp; image-segmentation; video-analysis; arabic-nlp", "keywords": "", "abstract": "Détection et reconnaissance du texte arabe incrusté dans les vidéos Cette thèse s'intéresse à la détection et la reconnaissance du texte arabe incrusté dans les vidéos. Dans ce contexte, nous proposons différents prototypes de détection et d'OCR vidéo (Optical Character Recognition) qui sont robustes à la complexité du texte arabe (différentes échelles, tailles, polices, etc.) ainsi qu'aux différents défis liés à l'environnement vidéo et aux conditions d'acquisitions (variabilité du fond, luminosité, contraste, faible résolution, etc.). Nous introduisons différents détecteurs de texte arabe qui se basent sur l'apprentissage artificiel sans aucun prétraitement. Les détecteurs se basent sur des Réseaux de Neurones à Convolution (ConvNet) ainsi que sur des schémas de boosting pour apprendre la sélection des caractéristiques textuelles manuellement conçus. Quant à notre méthodologie d'OCR, elle se passe de la segmentation en traitant chaque image de texte en tant que séquence de caractéristiques grâce à un processus de scanning. Contrairement aux méthodes existantes qui se basent sur des caractéristiques manuellement conçues, nous proposons des représentations pertinentes apprises automatiquement à partir des données. Nous utilisons différents modèles d'apprentissage profond, regroupant des Auto-Encodeurs, des ConvNets et un modèle d'apprentissage non-supervisé, qui génèrent automatiquement ces caractéristiques. Chaque modèle résulte en un système d'OCR bien spécifique. Le processus de reconnaissance se base sur une approche connexionniste récurrente pour l'apprentissage de l'étiquetage des séquences de caractéristiques sans aucune segmentation préalable. Nos modèles d'OCR proposés sont comparés à d'autres modèles qui se basent sur des caractéristiques manuellement conçues. Nous proposons, en outre, d'intégrer des modèles de langage (LM) arabes afin d'améliorer les résultats de reconnaissance. Nous introduisons différents LMs à base des Réseaux de Neurones Récurrents capables d'apprendre des longues interdépendances linguistiques. Nous proposons un schéma de décodage conjoint qui intègre les inférences du LM en parallèle avec celles de l'OCR tout en introduisant un ensemble d’hyper-paramètres afin d'améliorer la reconnaissance et réduire le temps de réponse. Afin de surpasser le manque de corpus textuels arabes issus de contenus multimédia, nous mettons au point de nouveaux corpus manuellement annotés à partir des flux TV arabes. Le corpus conçu pour l'OCR, nommé ALIF et composée de 6,532 images de texte annotées, a été publié a des fins de recherche. Nos systèmes ont été développés et évalués sur ces corpus. L’étude des résultats a permis de valider nos approches et de montrer leurs efficacité et généricité avec plus de 97% en taux de détection, 88.63% en taux de reconnaissance mots sur le corpus ALIF dépassant ainsi un des systèmes d'OCR commerciaux les mieux connus par 36 points.", "source_url": "https:\/\/openalex.org\/W7147073477", "pdf_url": "https:\/\/hal.science\/tel-01406716v2\/document", "doi": "10.70675\/531c973dzdb13z4a97z8f20zc8738a629707", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7147073477", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 140, "title": "Anomaly-based network intrusion detection using machine learning", "title_ar": "", "title_en": "", "author": "Maxime Labonne", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2020, "language": "fr", "research_domain": "machine-learning", "research_subdomain": "ai-cybersecurity", "categories": "machine-learning; ai-applications; ai-cybersecurity; ai-industry", "keywords": "", "abstract": "Détection d'intrusion réseau par anomalies avec apprentissage automatique Ces dernières années, le piratage est devenu une industrie à part entière, augmentant le nombre et la diversité des cyberattaques. Les menaces qui pèsent sur les réseaux informatiques vont des logiciels malveillants aux attaques par déni de service, en passant par le phishing et l'ingénierie sociale. Un plan de cybersécurité efficace ne peut plus reposer uniquement sur des antivirus et des pare-feux pour contrer ces menaces : il doit inclure plusieurs niveaux de défense. Les systèmes de détection d'intrusion (IDS) réseaux sont un moyen complémentaire de renforcer la sécurité, avec la possibilité de surveiller les paquets de la couche 2 (liaison) à la couche 7 (application) du modèle OSI. Les techniques de détection d'intrusion sont traditionnellement divisées en deux catégories : la détection par signatures et la détection par anomalies. La plupart des IDS utilisés aujourd'hui reposent sur la détection par signatures ; ils ne peuvent cependant détecter que des attaques connues. Les IDS utilisant la détection par anomalies sont capables de détecter des attaques inconnues, mais sont malheureusement moins précis, ce qui génère un grand nombre de fausses alertes. Dans ce contexte, la création d'IDS précis par anomalies est d'un intérêt majeur pour pouvoir identifier des attaques encore inconnues.Dans cette thèse, les modèles d'apprentissage automatique sont étudiés pour créer des IDS qui peuvent être déployés dans de véritables réseaux informatiques. Tout d'abord, une méthode d'optimisation en trois étapes est proposée pour améliorer la qualité de la détection : 1\/ augmentation des données pour rééquilibrer les jeux de données, 2\/ optimisation des paramètres pour améliorer les performances du modèle et 3\/ apprentissage ensembliste pour combiner les résultats des meilleurs modèles. Les flux détectés comme des attaques peuvent être analysés pour générer des signatures afin d'alimenter les bases de données d'IDS basées par signatures. Toutefois, cette méthode présente l'inconvénient d'exiger des jeux de données étiquetés, qui sont rarement disponibles dans des situations réelles. L'apprentissage par transfert est donc étudié afin d'entraîner des modèles d'apprentissage automatique sur de grands ensembles de données étiquetés, puis de les affiner sur le trafic normal du réseau à surveiller. Cette méthode présente également des défauts puisque les modèles apprennent à partir d'attaques déjà connues, et n'effectuent donc pas réellement de détection d'anomalies. C'est pourquoi une nouvelle solution basée sur l'apprentissage non supervisé est proposée. Elle utilise l'analyse de l'en-tête des protocoles réseau pour modéliser le comportement normal du trafic. Les anomalies détectées sont ensuite regroupées en attaques ou ignorées lorsqu'elles sont isolées. Enfin, la détection la congestion réseau est étudiée. Le taux d'utilisation de la bande passante entre les différents liens est prédit afin de corriger les problèmes avant qu'ils ne se produisent.", "source_url": "https:\/\/openalex.org\/W7149556586", "pdf_url": "https:\/\/theses.hal.science\/tel-02988296\/document", "doi": "10.70675\/7f7fbc77zaab9z4f60z9b41z7236bc47d138", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7149556586", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 153, "title": "Machine Learning Methods for Network Intrusion Detection and Intrusion Prevention Systems", "title_ar": "", "title_en": "Machine Learning Methods for Network Intrusion Detection and Intrusion Prevention Systems", "author": "Zheni Svetoslavova Stefanova", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of South Florida", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "responsible-ai", "research_subdomain": "fairness", "categories": "machine-learning; responsible-ai; ai-applications; predictive-analytics; fairness; ai-cybersecurity", "keywords": "", "abstract": "Given the continuing advancement of networking applications and our increased dependence upon software-based systems, there is a pressing need to develop improved security techniques for defending modern information technology (IT) systems from malicious cyber-attacks. Indeed, anyone can be impacted by such activities, including individuals, corporations, and governments. Furthermore, the sustained expansion of the network user base and its associated set of applications is also introducing additional vulnerabilities which can lead to criminal breaches and loss of critical data. As a result, the broader cybersecurity problem area has emerged as a significant concern, with many solution strategies being proposed for both intrusion detection and prevention. Now in general, the cybersecurity dilemma can be treated as a conflict-resolution setup entailing a security system and minimum of two decision agents with competing goals (e.g., the attacker and the defender). Namely, on the one hand, the defender is focused on guaranteeing that the system operates at or above an adequate (specified) level. Conversely, the attacker is focused on trying to interrupt or corrupt the system’s operation. In light of the above, this dissertation introduces novel methodologies to build appropriate strategies for system administrators (defenders). In particular, detailed mathematical models of security systems are developed to analyze overall performance and predict the likely behavior of the key decision makers influencing the protection structure. The initial objective here is to create a reliable intrusion detection mechanism to help identify malicious attacks at a very early stage, i.e., in order to minimize potentially critical consequences and damage to system privacy and stability. Furthermore, another key objective is also to develop effective intrusion prevention (response) mechanisms. Along these lines, a machine learning based solution framework is developed consisting of two modules. Specifically, the first module prepares the system for analysis and detects whether or not there is a cyber-attack. Meanwhile, the second module analyzes the type of the breach and formulates an adequate response. Namely, a decision agent is used in the latter module to investigate the environment and make appropriate decisions in the case of uncertainty. This agent starts by conducting its analysis in a completely unknown milieu but continually learns to adjust its decision making based upon the provided feedback. The overall system is designed to operate in an automated manner without any intervention from administrators or other cybersecurity personnel. Human input is essentially only required to modify some key model (system) parameters and settings. Overall, the framework developed in this dissertation provides a solid foundation from which to develop improved threat detection and protection mechanisms for static setups, with further extensibility for handling streaming data.", "source_url": "https:\/\/openalex.org\/W2912471702", "pdf_url": "https:\/\/digitalcommons.usf.edu\/context\/etd\/article\/8564\/viewcontent\/etd__7367.pdf", "doi": "", "handle": "", "repository_name": "Digital Commons - University of South Florida (University of South Florida)", "source_record_id": "W2912471702", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 161, "title": "Generative Quantum Machine Learning", "title_ar": "", "title_en": "Generative Quantum Machine Learning", "author": "Christa Zoufal", "supervisor": "", "degree_type": "phd", "degree_evidence": "نص السجل: «doctoral thesis»", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "generative-ai", "research_subdomain": "gan", "categories": "machine-learning; generative-ai; quantum-ml; gan", "keywords": "", "abstract": "The goal of generative machine learning is to model the probability distribution underlying a given data set. This probability distribution helps to characterize the generation process of the data samples. While classical generative machine learning is solely based on classical resources, generative quantum machine learning can also employ quantum resources - such as parameterized quantum channels and quantum operators - to learn and sample from the probability model of interest. Applications of generative (quantum) models are multifaceted. The trained model can generate new samples that are compatible with the given data and extend the data set. Additionally, learning a model for the generation process of a data set may provide interesting information about the corresponding properties. With the help of quantum resources, the respective generative models have access to functions that are difficult to evaluate with a classical computer and may improve the performance or lead to new insights. Furthermore, generative quantum machine learning can be applied to efficient, approximate loading of classical data into a quantum state which may help to avoid - potentially exponentially - expensive, exact quantum data loading. The aim of this doctoral thesis is to develop new generative quantum machine learning algorithms, demonstrate their feasibility, and analyze their performance. Additionally, we outline their potential application to efficient, approximate quantum data loading. More specifically, we introduce a quantum generative adversarial network and a quantum Boltzmann machine implementation, both of which can be realized with parameterized quantum circuits. These algorithms are compatible with first-generation quantum hardware and, thus, enable us to study proof of concept implementations not only with numerical quantum simulations but also real quantum hardware available today.", "source_url": "https:\/\/openalex.org\/W3212842386", "pdf_url": "https:\/\/arxiv.org\/pdf\/2111.12738", "doi": "10.3929\/ethz-b-000514692", "handle": "", "repository_name": "arXiv (Cornell University)", "source_record_id": "W3212842386", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 164, "title": "Quantum Machine Learning Applications and Algorithms", "title_ar": "", "title_en": "", "author": "Marco Russo", "supervisor": "", "degree_type": "phd", "degree_evidence": "نص السجل: «Ph.D»", "university": "", "university_source": "", "college": "", "department": "", "country": "Portugal", "governorate": "", "publication_year": 2026, "language": "it", "research_domain": "machine-learning", "research_subdomain": "quantum-ml", "categories": "machine-learning; quantum-ml", "keywords": "", "abstract": "This dissertation is presented in partial fulfillment of the requirements for Ph.D. degree in the Graduate School of Politecnico di Torino (ScuDo).La scrittura scientifica è notoriamente fredda, austera, priva di emozioni.Qualsiasi risultato ottenuto da un esperimento, sia esso inconcludente o grandioso, viene presentato con la solita formula \"in questa figura e in quella tabella si può vedere che\", immancabilmente seguito da un \"possiamo quindi dedurre che\".Poi un abisso, oltre il quale c'è un altro mondo, quello dell'espressione dell'anima, dei nostri moti interiori e della consapevolezza di essi.Abbiamo deciso appunto di metterci un abisso in mezzo, e in fondo non è che si possa pretendere altrimenti; la scienza è questione di fatti e non di opinioni.Ma dentro colui che scrive con freddezza, nel momento in cui lo fa, il suo sentire non si spegne.Meglio allora non smettere mai di avere la testa nel secondo mondo anche se ci tocca lavorare nel primo.Il sentire di chi scrive si nasconde tra ogni riga.E se dal punto di vista dell'utile conta quello che nelle prossime pagine sarà nero su bianco, quel che importa a me sono in realtà le righe invisibili.Ognuna di queste racchiude chi sono mentre scrivo, e le persone e le cose che in qualche modo, in fasi diverse di questo percorso, hanno avuto un impatto di cui magari non si rendono conto.È per loro che quelle righe bianche sono scritte.Per mio fratello, in cui vedo molta più forza di quanta ne veda lui in sé stesso, e che, anche se ogni tanto la marea glielo nasconde e lui se lo dimentica, è così in gamba che alla fine si guarderà indietro ridendo.Per mia madre, che non ha mai potuto fare a meno di essere preoccupata per qualsiasi cosa, che ha sempre saputo ascoltarmi in ogni momento mettendo da parte ogni impegno, e che quando sono andato in posti incredibili sembrava vivere quelle cose attraverso i miei occhi.Una vita passata a sentire più di noi le nostre stesse emozioni, sognando con noi.", "source_url": "https:\/\/openalex.org\/W7217833618", "pdf_url": "https:\/\/iris.polito.it\/bitstream\/11583\/3012489\/2\/phd_thesis.pdf", "doi": "", "handle": "", "repository_name": "PORTO@Iris", "source_record_id": "W7217833618", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 186, "title": "Spiking graph neural networks: A survey", "title_ar": "", "title_en": "Spiking graph neural networks: A survey", "author": "Sathvik Murli", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Rutgers University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "deep-learning", "research_subdomain": "cnn", "categories": "deep-learning; ai-applications; cnn; gnn; ai-cybersecurity; ai-smart-cities", "keywords": "", "abstract": "Spiking Neural Networks (SNNs) offer a biologically grounded alternative to conventional Artificial Neural Networks by replacing continuous activations with discrete, asynchronous spike events, while Graph Neural Networks (GNNs) provide a principled way to learn over non-Euclidean relational data. This paper surveys the emerging family of Spiking Graph Neural Networks (SGNNs), which unify these paradigms under a common modular template of a graph convolution backbone, a spike encoding stage, one or more Leaky Integrate-and-Fire (LIF) neuron layers, and a classification readout. Ten representative architectures are compared along four axes: the neuron model, the graph convolution mechanism, the treatment of time (static versus dynamic graphs), and the underlying representation space (Euclidean versus Riemannian). We further review training algorithms, showing that the non-differentiable spikefunction has driven three complementary solutions: surrogate-gradient Backpropagation Through Time, implicit or manifold-based differentiation with O(1) memory, and biologically-local Spike-Timing Dependent Plasticity. To demonstrate practical utility, we also present a case study applying an GNN inspired by SiGNN to malicious DNS-over-HTTPS traffic detection, where per-timestamp graphs are processed by graph convolutional layers followed by LIF-based spiking layers. The results indicate that SGNNs remain resilient to the strong class imbalance of the benchmark, supporting their viability for cybersecurity classification. Taken together, the surveyed work suggests that spiking graph models can inherit the energy benefits of neuromorphic hardware without sacrificing the expressive power of modern graph learning.", "source_url": "https:\/\/openalex.org\/W7165632984", "pdf_url": "", "doi": "10.7282\/t3-1wzw-gs74", "handle": "", "repository_name": "Rutgers University Community Repository (Rutgers University)", "source_record_id": "W7165632984", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 191, "title": "Learning latent graph structures for graph neural networks", "title_ar": "", "title_en": "Learning latent graph structures for graph neural networks", "author": "Haicang Zhou", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Nanyang Technological University", "university_source": "repository_host", "college": "", "department": "", "country": "Singapore", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; self-semi-supervised; data-mining; transformers; gnn", "keywords": "", "abstract": "Graphs are powerful tools for modeling complex relationships between entities, making them essential in various domains, such as social networks, citation networks, molecular graphs, and road networks. In recent years, Graph Neural Networks (GNNs) have gained significant attention for learning on graphs. However, existing GNNs are very limited in effectively learning various latent graph structures, which are not immediately apparent but very important for graph learning, such as graph clusters. Consequently, there is a pressing need to design novel approaches that enable GNNs to capture diverse latent structures in graph learning. To this end, this dissertation aims to design novel methodologies that can integrate latent structures into graph neural networks to bolster their learning performance. In particular, the focus is on three types of latent graph structures: graph clusters, implicit sparsity and latent structure from the Third Law of Geography. With our investigation on these structures and method designs, we can obtain more insights on both graph structures and graph learning, and further improve the performance of graph neural networks and their real-world applications. First, we highlight the importance of graph clusters in graph learning, and identify the benefits of incorporating graph clusters into graph attention which lacks cluster information. To incorporate graph clusters into graph attention, we proposed Differentiable Clustering for Graph Attention (DCAT), which enhances graph attention by learning cluster-aware attention scores. In DCAT, graph clusters are learned through a differentiable objective based on a relaxed form of modularity maximization. Thus the objective can be optimized together with the semi-supervised loss in an end-to-end manner. Our analysis indicates that DCAT can allocate higher attention scores to nodes within the same cluster, allowing them to have a higher influence on node representation learning. Second, we investigate the latent sparsity phenomenon in real-world graphs, where a large portion of edges connect irrelevant nodes, potentially degrading GNN performance. To address this issue, we propose a novel graph selective attention mechanism,which employs various forms of learnable node-node dissimilarity to learn the latent sparsity. With selective attention, each node can effectively filter out irrelevant edges and selectively attend to its most irrelevant neighbors. We further build the graph selective attention network (SAT) based on it, and conduct theoretical analysis on its expressiveness. Third, we investigate a latent structure in road networks according to the Third Law of Geography, which has not been explored in previous literature. This law suggests a latent structure which goes beyond the spatial relationships of geospatial entities according to their geographic configurations. To learn from the latent structure, we propose a novel graph contrastive learning method, with geographic configuration aware graph augmentation and spectral negative sampling. Our proposed approaches have been evaluated against a number of state-of-the-art baselines across various real-world tasks. The results demonstrate our methods can perform robustly, showcasing learning latent graph structures is indeed an effective way to advance graph neural networks. The methodologies proposed in this thesis inaugurate new directions that can further strengthen graph representation learning for various real-world tasks.", "source_url": "https:\/\/openalex.org\/W4410047771", "pdf_url": "", "doi": "10.32657\/10356\/184288", "handle": "", "repository_name": "DR-NTU (Nanyang Technological University)", "source_record_id": "W4410047771", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 194, "title": "Explaining Graph Neural Networks", "title_ar": "", "title_en": "Explaining Graph Neural Networks", "author": "Tobias Christian Nauen", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Leibniz Universität Hannover", "university_source": "repository_host", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "computer-vision", "research_subdomain": "gnn", "categories": "machine-learning; deep-learning; computer-vision; responsible-ai; predictive-analytics; gnn; xai", "keywords": "", "abstract": "Graph Neural Networks are an up-and-coming class of neural networks that operate on graphs and can therefore deal with connected, highly complex data. As explaining neural networks becomes more and more important, we investigate different ways to explain graph neural networks and contrast gradient-based explanations with the interpretability by design approach KEdge. We extend KEdge, to work with probability distributions different from HardKuma. Our goal is to test the performance of each method to judge which one works best under given circum- stances. For this, we extend the notion of fidelity from hard attribution weights to soft attribution weights and use the resulting metric to evaluate the explanations generated by KEdge, as well as by the gradient-based techniques. We also compare the predictive performance of models that use KEdge with different distributions. Our experiments are run on the Cora, SightSeer, Pubmed, and MUTAG datasets. We find that KEdge outperforms the gradient based attribution techniques on graph classification problems and that it should be used with the HardNormal, HardKuma, or HardLaplace distributions, depending on if the top priority is model performance or attribution quality. To compare different metrics of judging attributions in the text domain, we visualize attribution weights generated by different models and find, that metrics which compare model attributions to human explanations lead to bad attribution weights.", "source_url": "https:\/\/openalex.org\/W3217520584", "pdf_url": "", "doi": "10.15488\/11525", "handle": "", "repository_name": "Institutional Repository of Leibniz Universität Hannover (Leibniz Universität Hannover)", "source_record_id": "W3217520584", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 212, "title": "Developing and Evaluating Fairness-Aware Machine Learning Methods; Ethical Responsibility and Accountability for Algorithmic Bias in Hiring", "title_ar": "", "title_en": "Developing and Evaluating Fairness-Aware Machine Learning Methods; Ethical Responsibility and Accountability for Algorithmic Bias in Hiring", "author": "Tim Wang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "computer-vision", "research_subdomain": "face-recognition", "categories": "machine-learning; computer-vision; responsible-ai; predictive-analytics; face-recognition; video-analysis; fairness", "keywords": "", "abstract": "Introduction Machine learning systems have increasingly taken control of making high-stake decisions about people’s lives. This can include who advances in a hiring pipeline, who is released before trial, or whose face a security camera flag. The reason behind this mass adoption is the efficiency and apparent objectivity that machine learning systems bring. What gets left out is that this objectivity that is spoken of by organizations is often a fiction. When training a model using historical data, it is often shaped by the discrimination apparent within this data. So instead of correcting them, it encodes and automates these patterns. Amazon’s recruiting algorithm penalized resumes containing the word “women’s” as it had learned from a male-dominated workforce. HireVue’s video interviewing system was responsible for generating disparate outcomes across demographic groups and even faced EEOC charges in 2025. These are the inevitable outcomes that come from machine learning systems optimizing for accuracy with no questions raised about fairness. My thesis addresses this problem with two directions: A technical project proposing a framework for evaluating fairness-aware machine learning methods, and an STS project that uses Actor-Network Theory to analyze how ethical responsibility for discriminatory hiring algorithms is distributed and evaded. Technical Project The technical portion of my thesis proposes a systematic framework for comparing bias mitigation techniques across three pipeline stages, which are preprocessing, in-processing, and postprocessing. It includes the use of three datasets: The Adult Income dataset, a synthetic pretrial risk assessment dataset that will be constructed from public criminal justice statistics, and facial recognition benchmarks. The synthetic dataset is critical to the project as it allows control over what known bias patterns are explicitly implemented into the data and therefore the framework can test whether each method removed them, which is impossible with proprietary tools like COMPAS. The evaluation will use Python, scikit-learn, and IBM’s AI Fairness 360 toolkit to measure not only the predictive performance but also fairness metrics which will include demographic parity, equalized odds, and equalized opportunity. The expected result from the project is a practical and empirically grounded understanding of the tradeoffs that come from each type of method and in which domains each of them work best. STS Research Paper In my STS research, I applied Actor-Network Theory to discover and trace how responsibility for biased hiring algorithms is formed and deflected. ANT treats both human and non-human actors symmetrically, so executives, data scientists, training datasets, and legal frameworks are all actants that actively shape the outcomes; therefore blame cannot rest at any single node. Through case studies of Amazon’s recruiting tool, HireVue’s video system, and the 2024 Mobley v. Workday lawsuit, I have mapped how accountability spreads when no one actor has full control of the system. The employers will blame vendors, vendors claim they only built the tool, and regulators find civil rights laws ill-suited to algorithmic decision-making. Drawing from the works of O’Neil, Noble, and Eubanks, my research argues this diffusion is a structural feature of how these networks are formed and recommends three specific interventions: mandatory algorithmic impact assessments, enforceable third-party bias audits with the inclusion of public disclosure, and vendor liability reforms. Conclusion When put together, these two projects illustrate a core tension as the technical work demonstrates that fairness interventions are feasible, while the STS research shows that this technical feasibility will not matter if there are no accountability structures that require their use. The technical framework strengthens the argument that developers who create the algorithms and employers who use them have an obligation to act. The STS analysis details why these tools remain underused, as organizations have very little reason to adopt these fairness-aware methods when the alternative carries no real legal or reputational risk. Hiring is a primary domain where this often occurs in and automating discrimination at scale behind the façade of objectivity causes harm that is harder to detect and contest than its human equivalent. Addressing this issue requires both the technical capability of building fairer systems and the institutional will to require that organizations use them.", "source_url": "https:\/\/openalex.org\/W7159615210", "pdf_url": "", "doi": "10.18130\/svmp-4c59", "handle": "", "repository_name": "Libra", "source_record_id": "W7159615210", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 216, "title": "A Comparative Benchmark of Fairness Metrics in Machine Learning", "title_ar": "", "title_en": "A Comparative Benchmark of Fairness Metrics in Machine Learning", "author": "Md Asif Shahariar", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Tampere University", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "responsible-ai", "research_subdomain": "xai", "categories": "machine-learning; responsible-ai; predictive-analytics; xai; fairness", "keywords": "", "abstract": "Machine learning is no longer limited to low-risk tasks, but it is also being used in critical decision-making systems. Consequently, machine learning has a direct effect on individuals' lives in domains such as recidivism prediction (COMPAS) and loan approval. Thus, errors or biases generated by such algorithms may have severe consequences in the real world. Therefore, our expectations of ML are not limited to accuracy. In addition to making correct predictions, we expect the models to behave in an unbiased manner toward individuals and groups. Many mathematical definitions of fairness have been proposed by researchers, but there have been few large, systematic comparisons of fairness metrics applied to standard machine learning models. The main task of this thesis is to examine in detail how the balance or trade-off between accuracy and fairness occurs in different models. The study used two socially sensitive datasets, COMPAS recidivism and Adult Income. These two datasets were used to train and test nine machine learning models. The models used in this study include nine widely used binary classifiers, ranging from simple and interpretable algorithms (e.g., Logistic Regression and Support Vector Machine) to high-performance ensemble methods (e.g., Random Forest, XGBoost, and LightGBM). Each model was carefully evaluated using Demographic Parity Difference (DPD) and Equalized Odds Difference (EOD) which are two important group fairness criteria, and traditional performance metrics including Accuracy and F1-score. The outcome of the experiment clearly shows a trade-off between accuracy and fairness despite the fact that the nature of this relationship varies greatly among the model architectures. Although ensemble algorithms including XGBoost and LightGBM achieved the highest predictive accuracy, they also had the largest fairness gaps. On the other hand, less complex models like Decision Trees and Multi-layer Perceptrons tended to be more fair and less accurate. One of the key findings in the hyperparameter sensitivity analysis is the fact that fairness is not fixed. It varied with hyperparameters for models like Logistic Regression and SVM, but was less responsive for others. Furthermore, statistical significance testing highlighted that while performance advantages were often robust, differences in fairness rankings were frequently statistically insignificant, cautioning against over-reliance on minor metric variations. This thesis develops a robust empirical foundation for fairness assessment in machine learning through broad benchmarking with both sensitivity analysis and statistical testing. This study serves as a methodological guide and a source of empirical evidence for practitioners and researchers, providing a framework for evaluating fairness based on evidence-based insights. These results indicate that fairness is not universal or transferable across models or datasets, it must be evaluated contextually, with awareness of model behavior and tuning sensitivity.", "source_url": "https:\/\/openalex.org\/W7132936585", "pdf_url": "https:\/\/trepo.tuni.fi\/bitstream\/10024\/232398\/4\/ShahariarMdAsif.pdf", "doi": "", "handle": "", "repository_name": "Tampere University Institutional Repository (Tampere University)", "source_record_id": "W7132936585", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 220, "title": "RESPONSIBLE RISK ASSESSMENT OF MACHINE LEARNING TOOLS FOR PROMOTION BOARDS", "title_ar": "", "title_en": "RESPONSIBLE RISK ASSESSMENT OF MACHINE LEARNING TOOLS FOR PROMOTION BOARDS", "author": "Michael R. Burke", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Naval Postgraduate School", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "responsible-ai", "research_subdomain": "fairness", "categories": "machine-learning; robotics; responsible-ai; predictive-analytics; fairness", "keywords": "", "abstract": "This thesis evaluates an artificial intelligence (AI) model's ability to support U.S. Marine Corps promotion boards, investigating how data drift and feedback loops impact its performance and reliability over time. Using the DataRobot automated machine learning (AutoML) platform, a model was trained on fiscal year (FY) 2023 Major promotion board data and subsequently tested against FY24 and FY25 data to measure performance degradation and detect statistical drift in key predictors. The model demonstrated strong but brittle performance, maintaining high ranking accuracy across years while relying heavily on administrative features like promotion zone, which exhibited significant data drift. The research concludes that without robust governance, the model's dependence on procedural proxies risks amplifying biases and capturing administrative timing rather than a holistic view of merit, potentially undermining the promotion system's integrity. It is recommended that the Marine Corps use the AI tool for pre-screening to improve efficiency, while implementing annual retraining, continuous drift monitoring, and explicit fairness thresholds to ensure long-term effectiveness and trust.", "source_url": "https:\/\/openalex.org\/W7134863984", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Calhoun: The Naval Postgraduate School Institutional Archive (Naval Postgraduate School)", "source_record_id": "W7134863984", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 222, "title": "Towards Trustworthy Machine Learning: Fairness Under Dynamics and Counterfactuals", "title_ar": "", "title_en": "Towards Trustworthy Machine Learning: Fairness Under Dynamics and Counterfactuals", "author": "Ozgur Guldogan", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; nlp; generative-ai; responsible-ai; few-zero-shot; predictive-analytics; llm; fairness", "keywords": "", "abstract": "Machine learning systems increasingly make decisions with long-term consequences for individuals and society. Designing trustworthy machine learning systems therefore requires going beyond static accuracy and one-shot fairness criteria. Decisions may alter future populations through feedback, individuals may improve and reapply, uncertainty may call for set-valued predictions rather than point estimates, and large-scale deployment may be constrained by the efficiency of modern generative models. This thesis explores these challenges through contributions spanning fairness under dynamics, counterfactual fairness under predictive uncertainty, and efficient inference serving for large language models. We first study fairness in dynamic decision-making settings such as lending, hiring, and admissions, where rejected individuals may invest effort to improve and reapply. We introduce Equal Improvability, a fairness notion requiring that rejected individuals from different groups have comparable opportunities to become accepted after bounded effort. We develop optimization-based methods for learning classifiers that satisfy this criterion and show that they can promote more equitable long-term population dynamics. Next, we consider sequential selection by multiple decision-makers from a shared applicant pool. We propose fair-greedy policies that balance utility with long-term representation targets, characterize convergence under positive reinforcement dynamics, and show that policy design must account for more complex feedback mechanisms to avoid unintended negative reinforcement. We then address fairness under counterfactual reasoning and predictive uncertainty. Counterfactually Fair Conformal Prediction constructs prediction sets that are both statistically valid and counterfactually fair. By aggregating conformity scores across counterfactual interventions on protected attributes, the method produces prediction sets that remain invariant under counterfactual changes to the protected attribute while preserving finite-sample marginal coverage guarantees. Finally, we study computational efficiency in generative machine learning systems. Multi-Bin Batching is an inference-serving policy for large language models that groups requests by predicted generation lengths to reduce inefficiency from heterogeneous runtimes. Queueing-theoretic analysis characterizes the throughput benefits of length-aware batching and shows that the effective service rate approaches the idealized static-batching capacity as the number of bins increases. Experiments on simulated and real inference workloads demonstrate substantial throughput improvements over static batching. Together, these contributions provide fairness notions and algorithms that remain meaningful under feedback, uncertainty, and counterfactual interventions, while also developing principled serving mechanisms that support scalable deployment of generative models.", "source_url": "https:\/\/openalex.org\/W7213476610", "pdf_url": "https:\/\/escholarship.org\/content\/qt0r1882jr\/qt0r1882jr.pdf", "doi": "", "handle": "", "repository_name": "eScholarship (California Digital Library)", "source_record_id": "W7213476610", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 225, "title": "Sensitivity analysis and algorithmic fairness for machine learning and artificial intelligence", "title_ar": "", "title_en": "", "author": "Clément Bénesse", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2022, "language": "fr", "research_domain": "responsible-ai", "research_subdomain": "fairness", "categories": "machine-learning; responsible-ai; fairness", "keywords": "", "abstract": "Analyse de sensibilité et équité algorithmique pour le machine learning et l'intelligence artificielle Au cours des dernières années, le déploiement d'algorithmes utilisant du Machine Learning a explosé. En permanence à la recherche de performances améliorées au prix de toujours plus de données, ces algorithmes ont révolutionné de nombreux domaines. Néanmoins, de plus en plus de problèmes transparaissent lors de l'utilisation de ces modèles. Dans cette thèse, nous nous intéressons à deux écueils particuliers que sont le manque d'explicabilité, ainsi que le défaut d'équité pouvant mener à des discriminations qui peut être présent dans ces algorithmes. Nous montrons comment ces deux sujets sont liés en présentant un cadre probabiliste unificateur permettant de transposer des techniques de l'un des domaines vers l'autre.", "source_url": "https:\/\/openalex.org\/W7151539384", "pdf_url": "https:\/\/utheme.utoulouse.fr\/files\/original\/dbbe0ed98fc2d0f31e153387c81e933fa6b09906.pdf", "doi": "10.70675\/0445eb0az81f4z4ea8z8a87z1a0fef986a20", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7151539384", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 234, "title": "Optimizing Operating Costs at EMX Using Machine Learning-Based Forecasting And Capacity Planning Models", "title_ar": "", "title_en": "Optimizing Operating Costs at EMX Using Machine Learning-Based Forecasting And Capacity Planning Models", "author": "Sara Hussein Abu Iktish", "supervisor": "", "degree_type": "master", "degree_evidence": "نص السجل: «Master»", "university": "American University of Sharjah", "university_source": "repository_host", "college": "", "department": "", "country": "United Arab Emirates", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "machine-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; predictive-analytics", "keywords": "", "abstract": "A Master of Science thesis in Engineering Systems Management by Sara Hussein Abu Iktish entitled, “Optimizing Operating Costs at EMX Using Machine Learning-Based Forecasting and Capacity Planning Models”, submitted in July 2025. Thesis advisor is Dr. Hussam AlShraideh. Soft copy is available (Thesis, Completion Certificate, Approval Signatures, and AUS Archives Consent Form).", "source_url": "https:\/\/openalex.org\/W7162362814", "pdf_url": "", "doi": "10.82366\/411", "handle": "", "repository_name": "DSpace at AUS (American University of Sharjah)", "source_record_id": "W7162362814", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 252, "title": "Benchmarking machine learning models at the edge", "title_ar": "", "title_en": "Benchmarking machine learning models at the edge", "author": "Md. Masud Haider", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Tampere University", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "computer-vision", "research_subdomain": "", "categories": "machine-learning; computer-vision", "keywords": "", "abstract": "Recently the use of resource-constrained edge devices to perform machine learning tasks has gained substantial research attention. The need to perform such tasks arises from the vast amounts of sensors, and IoT and smart devices that generate large swaths of data at an enormous rate. To process such data in a real-time manner, cloud computing has proven to be suboptimal. As an alternative, edge devices paired with ML models can execute demanding tasks in diverse environments. This study explored a range of machine learning models coupled with several low and high-end edge devices to test their suitability for deployment in an edge environment. Unlike previous works, this experiment takes into account the initial heavy lifting that machine learning models and devices go through before performing the actual task of classification or detection. To systematically benchmark and gather results, this study divides the ML pipeline into initial and execution phases while thoroughly measuring each phase’s time and resource usage. This type of rigorous performance metric analysis enables decision-making about which machine learning model and devices have better suitability for latency and speed. Moreover, to add a device in the suit of existing ones and replicate the benchmark workflow, this study has automated the benchmark with the Ansible automation tool. The thesis research presents its findings from different devices. The division of phases allows for a clear understanding of the bottlenecks experienced by the devices and their capabilities. Moreover, an underlying crucial detail such as the warmup was revealed by scrutinizing the performances of the devices from the very beginning. The thesis concludes with performance highlights and limitations of the benchmark, identifying possible future research directions that can be developed further.", "source_url": "https:\/\/openalex.org\/W7133090547", "pdf_url": "https:\/\/trepo.tuni.fi\/bitstream\/10024\/161905\/2\/HaiderMd.pdf", "doi": "", "handle": "", "repository_name": "Tampere University Institutional Repository (Tampere University)", "source_record_id": "W7133090547", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 256, "title": "Comparative Performance Analysis of Deep Learning Models for Brain Tumor Classification from MRI Images", "title_ar": "", "title_en": "Comparative Performance Analysis of Deep Learning Models for Brain Tumor Classification from MRI Images", "author": "Hassan Khan", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Eastern Finland", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; transfer-learning; cnn; transformers; medical-imaging", "keywords": "", "abstract": "Brain tumors are among the most serious categories of neurological disease, with accurate classification from magnetic resonance imaging (MRI) playing a critical role in treatment planning and patient outcomes.Manual interpretation of MRI scans by radiologists is timeconsuming, subject to inter-observer variability, and constrained by the availability of specialist expertise, particularly in resource-limited settings.These factors have driven growing interest in automated deep learning solutions for brain tumor MRI classification.While convolutional neural networks (CNNs) have demonstrated strong performance on this task through transfer learning, the emergence of Vision Transformers and hybrid CNN-Transformer architectures has created a diverse model landscape whose relative merits remain poorly understood.Existing comparative studies benchmark either multiple CNN variants in isolation or transformer-based models separately, without unified head-tohead evaluation across all three architectural families under identical experimental conditions.Furthermore, deployment-relevant efficiency metrics -inference latency, model size, and GPU memory -are rarely reported alongside accuracy, limiting practical model selection guidance for clinical settings with varying hardware constraints.This thesis addresses these gaps by conducting a systematic benchmark of nine deep learning architectures -four CNN-based (VGG16, ResNet50, DenseNet121, EfficientNet-B0), three Transformer-based (ViT-B\/16, Swin Transformer, DeiT-B), and two Hybrid\/Lightweight (MobileNetV2, MaxViT-Tiny) -evaluated on the Brain Tumor MRI Dataset comprising 7,200 T1-weighted MRI images across four classes (glioma, meningioma, pituitary tumor, no tumor).All models were trained under identical conditions using ImageNet transfer learning and assessed on both classification performance (accuracy, precision, recall, F1-score, AUC-ROC) and computational efficiency (parameter count, model size, inference latency, peak GPU memory).The Swin Transformer achieves the highest test accuracy (98.52%,F1 = 0.9852), demonstrating that hierarchical transformer architectures can outperform CNNs when appropriately designed for the task.MobileNetV2 leads all models on efficiency metrics with 2.2M parameters and 7.55 ms inference latency.Pareto analysis identifies three deployment-optimal models: Swin-T for high-resource GPU environments, DenseNet121 (98.15%, 7.0M parameters) for standard workstations, and MobileNetV2 (97.04%) for edge devices.Efficiency varies by a factor of 61× in parameter count and 25× in GPU memory across the nine models, while accuracy spans only ~2 percentage points, underscoring that architectural design matters far more than raw model size.This study provides the first unified benchmark comparing CNN, Transformer, and Hybrid architectures for brain tumor MRI classification under identical conditions with both accuracy and efficiency metrics reported, yielding evidence-based model selection guidelines for clinical deployment across diverse computational environments.", "source_url": "https:\/\/openalex.org\/W7215551487", "pdf_url": "https:\/\/erepo.uef.fi\/bitstreams\/57ebe372-c179-491f-bae8-4ad920908485\/download", "doi": "", "handle": "", "repository_name": "UEF eRepo (University of Eastern Finland)", "source_record_id": "W7215551487", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 259, "title": "Advancements in Automated Brain Tumor Detection Using Deep Learning on MRI Imagery", "title_ar": "", "title_en": "Advancements in Automated Brain Tumor Detection Using Deep Learning on MRI Imagery", "author": "Md. Sohan Biswas", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; transfer-learning; cnn; medical-imaging", "keywords": "Brain tumor, Convolutional Neural Networks, Deep Learning, Transfer Learning", "abstract": "The goal of this thesis is to solve the significant problem of inter-observer variability-induced diagnostic delays by presenting a sophisticated deep learning system for automated brain tumor diagnosis utilizing MRI imagery. Pituitary adenoma, extra-axial meningioma, infiltrative glioma, and no-tumor controls are the four classes of brain tumors that are represented in the 7,143 T1 weighted MRI images taken from Kaggle Brain Tumor Dataset and various hospitals in Bangladesh, including Ibn Sina Medical College, Dhaka Medical College, and Comilla Medical College. Using three pretrained convolutional neural network (CNN) architectures; MobileNetV2, EfficientNetB0, and VGG16 optimized through selective layer unfreezing and on-the-fly data augmentation, including random brightness and contrast adjustments to mitigate scanner artifacts, the study applies transfer learning to improve diagnostic accuracy. With a test accuracy of 98.5%, a macro F1-score of 0.98, and an area under the curve (AUC) ≥0.99, MobileNetV2 outperformed the other models after being trained on 5,832 photos. Particularly noteworthy was the MobileNetV2 model's real-time inference ability (0.97) and quick processing times, especially in low-resource settings where prompt diagnosis can have a major impact on survival outcomes. The study highlights the clinical viability of applying AI-assisted diagnostics in various contexts, enhancing access to superior radiological evaluations and possibly cutting down on diagnostic delays in practical applications. Keywords: Brain tumor, Convolutional Neural Networks, Deep Learning, Transfer Learning, MRI Classification Source Code:GitHub: Advancements in Automated Brain Tumor Detection Using Deep Learning on MRI Imagery Author: Md. Sohan BiswasORCID: 0009-0000-5888-373X", "source_url": "https:\/\/openalex.org\/W7201959393", "pdf_url": "", "doi": "10.5281\/zenodo.21845324", "handle": "", "repository_name": "Zenodo (CERN European Organization for Nuclear Research)", "source_record_id": "W7201959393", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 262, "title": "Image analysis and machine learning in patients with brain tumors", "title_ar": "", "title_en": "Image analysis and machine learning in patients with brain tumors", "author": "Alice Neimantaite", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Gothenburg University", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; predictive-analytics; multimodal; medical-imaging", "keywords": "", "abstract": "Diffuse lower-grade gliomas (dLGG) are slowly growing primary brain tumors. Patients with dLGG represent a heterogeneous group with various symptoms, diverse progression of the disease as well as survival length. Collecting multidimensional patient data gives the opportunity for application of advanced analysis methods. These methods have shown a great promise with the ability to recognize complex patterns in the data and the potential to improve patient treatment and knowledge on the disease. However, the methods should be applied with consideration to the clinical context. This thesis aims to apply advanced analysis methods for the patients with dLGG in a clinically meaningful manner. Depending on the research question, patients with suspected or histomolecularly confirmed dLGG diagnosis from Sahlgrenska University hospital and external sites were included in the studies of the thesis. The analysed data included multi-modal MRI, clinical data, neuropsychological data and health-related quality of life (HRQoL) assessments. Radiomics was applied for survival prediction (Study I), deep learning for tumor classification (Study II), correlation analysis for investigation of symptom interconnection (Study III) and lesion mapping to find brain structures at risk in isocitrate dehydrogenase (IDH)- mutated gliomas (Study IV). The results showed that radiomics can identify clinically relevant information and a model combining radiomics features with clinical variables improved survival prediction (Study I). Deep learning showed similar performance as the clinical model in classification of IDH-mutation status in non-enhancing gliomas (Study II). Correlation networks revealed that self-reported and observer-assessed variables on mental fatigue and cognitive functioning were not connected (Study III). Lesion mapping quantified brain structures at risk in IDH-mutated gliomas, which were often related to advanced brain functions (Study IV). Advanced analysis methods have aided in the understanding of dLGG in several perspectives from symptoms to prognostic factors for patient survival. The machine learning model showed better performance than clinical models in survival prediction, but not in clinically relevant tumor classification. Our findings strengthen the notion of the importance to apply these methods in a relevant context to gain insight where the methods are clinically useful.", "source_url": "https:\/\/openalex.org\/W7217647342", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Gothenburg University Publications Electronic Archive (Gothenburg University)", "source_record_id": "W7217647342", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 266, "title": "Brain Tumor Segmentation using Deep Learning: A Hybrid Model.", "title_ar": "", "title_en": "Brain Tumor Segmentation using Deep Learning: A Hybrid Model.", "author": "Gishene, Tigabu Yaya", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Helsinki", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; medical-ai; image-segmentation; medical-imaging", "keywords": "", "abstract": "Accurate brain tumor segmentation from MRI scans is crucial for diagnosis, treatment planning, and prognosis. However, manual segmentation is labor-intensive, prone to variability, and susceptible to errors, emphasizing the need for automated solutions. This study proposes the development and evaluation of deep learning architectures for brain tumor segmentation using the BraTS 2020 dataset. We propose and compare two approaches: (1) a standalone 3D U-Net, which processes 3D MRI volumes, and (2) a hybrid 2D model that combines a pre-trained VGG19 encoder with a U-Net decoder for 2D slice-based segmentation. The hybrid model leverages VGG19’s powerful feature extraction capabilities while maintaining the structural advantages of U-Net. To optimize the hybrid model, we applied a data transformation strategy that transforms volumetric MRI data into 2D slices, aligning with the inherent architecture of VGG19. The segmentation models were trained and evaluated using three out of four MRI modalities from the BraTS 2020 dataset, T1Gd, T2, and FLAIR, which provide complementary information for accurate tumor structure identification. Performance evaluation showed that the 3D U-Net exhibits limitations in segmenting tumor subregions, achieving Dice Similarity Coefficients (DSC) of 0.7870 (Necrotic and Non-Enhancing Tumor Core, NCR\/NET), 0.8609 (Edema, ED), and 0.8058 (Enhancing Tumor, ET), with corresponding Intersection over Union (IoU) scores of 0.6489, 0.7558, and 0.6747, respectively. While the model demonstrates high specificity across all tumor classes, its lower sensitivity suggests the need for improvements to detect all instances of pathological tissue more effectively. On the other hand, the hybrid U-Net + VGG19 model demonstrates clearly superior segmentation results, outperforming many existing methods. It achieved DSC scores of 0.9078 (NCR\/NET), 0.9245 (ED), and 0.9175 (ET), consistently exceeding 0.91 across key tumor subregions. Additionally, the hybrid model achieves superior segmentation performance across the three tumor structures, whole tumor (WT), tumor core (TC), and enhancing tumor (ET), achieving DSC scores of 0.9620, 0.9550, and 0.9175, respectively. As observed from the performance evaluation results of the models, the hybrid U-Net with VGG19 provides a computationally efficient and highly accurate approach for brain tumor segmentation.", "source_url": "https:\/\/openalex.org\/W7115224021", "pdf_url": "https:\/\/www.doria.fi\/bitstream\/10024\/192488\/2\/gishene_tigabu_yaya.pdf", "doi": "", "handle": "", "repository_name": "Doria (University of Helsinki)", "source_record_id": "W7115224021", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 273, "title": "Characterization of Pseudo CT From MRI With Deep Learning : Application to Brain Tumors Treated With Radiotherapy", "title_ar": "", "title_en": "", "author": "Émilie Alvarez Andres", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2021, "language": "fr", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; medical-ai; image-segmentation; medical-imaging", "keywords": "", "abstract": "Caractérisation de pseudo CT générés à partir d'images IRM à l'aide de méthodes deep learning : application aux tumeurs cérébrales traitées par radiothérapie De nos jours, les traitements de tumeurs cérébrales par radiothérapie nécessitent l’acquisition d’un scanner utilisé pendant les étapes de segmentation et de dosimétrie, ainsi que d’une Imagerie par Résonance Magnétique (IRM) jouant un rôle important durant l’étape de segmentation des volumes cibles en particulier. Cependant, cette double modalité implique de recaler spatialement les images, processus qui induit des erreurs de 2mm, actuellement prises en compte par une augmentation de marges. Ainsi, générer des pseudo scanners (pCT) à partir d’images IRM apparaît comme étant une solution attractive pour diminuer les marges et réduire l’irradiation des tissus sains périphériques. La première étape de cette thèse avait pour but de caractériser les paramètres jouant un rôle clé dans la qualité de pCT générés par Deep Learning (DL), à savoir la taille de la cohorte d’entrainement, la séquence IRM utilisée en entrée du réseau, la technique de standardisation des images IRM, le filtre de correction d’inhomogénéités de champ et l’architecture du réseau. Pour ce faire, une large cohorte composée de plus de 400 patients a été constituée, rassemblant des images de multiples appareils d’IRM et localisations tumorales, afin d’assurer la robustesse du modèle. Les pCT obtenus ont tout d’abord été évalués à l’aide de l’erreur absolue moyenne, basée sur les intensités. Des analyses dosimétriques ont ensuite été menées. Toutes les approches étudiées ont atteint des performances dosimétriques équivalentes, excepté pour la taille du jeu d’entrainement. Pour introduire une dosimétrie basée sur les pCT en pratique clinique et déployer une méthodologie de validation de la non-infériorité de la planification de traitement dans le cas d’une radiothérapie basée seulement sur IRM comparée à celle conventionnellement basée sur scanner, une deuxième étude visant à définir les métriques dosimétriques les plus adaptées à une évaluation de pCT sans biais a été réalisée. Leurs corrélations avec des métriques basées sur les intensités ont été calculées. Enfin, l’impact de scenarios simulant des erreurs extrêmes de pCT a été quantifié, basé sur les métriques optimales préalablement définies. Un nouveau jeu de test de 71 patients a été constitué, reflétant les localisations tumorales rencontrées en clinique et les modalités de traitement propres à notre centre. Les gamma index globaux et locaux pour le critère 1%\/1mm associés à des seuils de dose non-extrêmes ont montré leur pertinence pour la tâche d’évaluation de pCT cérébraux. De plus, les différences de métriques issues des histogrammes dose\/volume des volumes cibles et organes à risque doivent aussi être calculées car elles reflètent la performance dosimétrique du pCT pour chaque structure segmentée. Enfin, la génération de pCT a été appliquée à la radiothérapie basée sur protons, grâce à une collaboration avec le Centre de Protonthérapie d’Orsay. Le modèle de DL préalablement développé sur des patients adultes a été testé sur des patients pédiatriques, afin d’évaluer sa généralisabilité. Des performances cliniques satisfaisantes ont été atteintes, excepté pour quelques patients, pouvant potentiellement prouver la transférabilité du modèle évalué. Les travaux futurs comprennent une évaluation dosimétrique à plus grande échelle, avec la composition d’une cohorte de 198 enfants représentant 4 histologies différentes. Le but est de déterminer l’approche d’entrainement et de validation du réseau la plus efficace à l’aide de cartes de pouvoirs d’arrêt pour s’affranchir de l’influence des paramètres des scanners sur les images. Ainsi, une solide compréhension des points clés de la génération de pCT ainsi qu’une méthodologie de leur caractérisation ont été réalisées. Les recommandations en découlant ont le rôle clé de faciliter la quantification et l’interprétation de critères d’évaluation de la qualité de pCT dans le contexte de mise en place d’essais cliniques, tel que l’essai observationnel en cours GliopCT.", "source_url": "https:\/\/openalex.org\/W4392422560", "pdf_url": "https:\/\/theses.hal.science\/tel-03816279v1\/document", "doi": "10.70675\/82d46582z5167z4aaeza3e3z4bce727d5a2f", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W4392422560", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 275, "title": "Deep learning for brain tumor segmentation and survival prediction.", "title_ar": "", "title_en": "Deep learning for brain tumor segmentation and survival prediction.", "author": "Tirivangani Magadza", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of KwaZulu-Natal", "university_source": "repository_host", "college": "", "department": "", "country": "South Africa", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; predictive-analytics; multimodal; image-segmentation; medical-imaging", "keywords": "", "abstract": "A brain tumor is an abnormal growth of cells in the brain that multiplies uncontrolled. The death of people due to brain tumors has increased over the past few decades. Early diagnosis of brain tumors is essential in improving treatment possibilities and increasing the survival rate of patients. The life expectancy of patients with glioblastoma multiforme (GBM), the most malignant glioma, using the current standard of care is, on average, 14 months after diagnosis despite aggressive surgery, radiation, and chemotherapies. Despite considerable efforts in brain tumor segmentation research, patient diagnosis remains poor. Accurate segmentation of pathological regions may significantly impact treatment decisions, planning, and outcome monitoring. However, the large spatial and structural variability among brain tumors makes automatic segmentation a challenging problem, leaving brain tumor segmentation an open challenge that warrants further research endeavors. While several methods automatically segment brain tumors, deep learning methods are becoming widespread in medical imaging due to their resounding performance. However, the boost in performance comes at the cost of high computational complexity. Therefore, to improve the adoption rate of computer-assisted diagnosis in clinical setups, especially in developing countries, there is a need for more computational and memoryefficient models. In this research, using a few computational resources, we explore various techniques to develop deep learning models accurately for segmenting the different glioma sub-regions, namely the enhancing tumor, the tumor core, and the whole tumor. We quantitatively evaluate the performance of our proposed models against the state-of-the-art methods using magnetic resolution imaging (MRI) datasets provided by the Brain Tumor Segmentation (BraTS) Challenge. Lastly, we use segmentation labels produced by the segmentation task and MRI multimodal data to extract appropriate imaging\/radiomic features to train a deep learning model for overall patient survival prediction.", "source_url": "https:\/\/openalex.org\/W4399612389", "pdf_url": "https:\/\/researchspace.ukzn.ac.za\/bitstreams\/179a4807-2828-41c0-8d3f-771f55ce5b41\/download", "doi": "10.29086\/10413\/23075", "handle": "", "repository_name": "ResearchSpace (University of KwaZulu-Natal)", "source_record_id": "W4399612389", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 280, "title": "Classification of hate speech in social media using Machine Learning and Natural Language Processing", "title_ar": "", "title_en": "Classification of hate speech in social media using Machine Learning and Natural Language Processing", "author": "Sampath Reddy Kalwa", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "machine-learning; deep-learning; nlp; predictive-analytics; rnn; speech-recognition", "keywords": "", "abstract": "This study aims to apply Natural Language Processing (NLP) and develop Machine Learning algorithms for detecting hate speech in social media. The “Hate Speech and Offensive Language dataset” is used, which comprises 24,783 manually annotated English tweets labelled as hate speech, offensive language, or neither. Random Forest Classifier, RNN, LSTM, and Bi-LSTM models are evaluated for classifying hate speech and recommending effective strategies for minimising offensive language on social media platforms. Model Comparison helps to understand the best-fitted model in the context of predicting and classifying hate speech in social media platforms. All the objectives of this research are achieved through the entire data analysis, from data pre-processing to the evaluation of different ML techniques. More Neural Network models and hyperparameter-tuned ML-models would enhance the model performance in classifying hate speech in the future.", "source_url": "https:\/\/openalex.org\/W7167554688", "pdf_url": "https:\/\/norma.ncirl.ie\/9471\/1\/sampathreddykalwa.pdf", "doi": "", "handle": "", "repository_name": "NORMA", "source_record_id": "W7167554688", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 286, "title": "Machine Learning Models for Efficient and Robust Natural Language Processing", "title_ar": "", "title_en": "Machine Learning Models for Efficient and Robust Natural Language Processing", "author": "Emma Strubell", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Massachusetts Amherst", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; nlp; cnn; rnn; transformers", "keywords": "", "abstract": "Natural language processing (NLP) has come of age. For example, semantic role labeling (SRL), which automatically annotates sentences with a labeled graph representing who did what to whom, has in the past ten years seen nearly 40% reduction in error, bringing it to useful accuracy. As a result, a myriad of practitioners now want to deploy NLP systems on billions of documents across many domains. However, state-of-the-art NLP systems are typically not optimized for cross-domain robustness nor computational efficiency. In this dissertation I develop machine learning methods to facilitate fast and robust inference across many common NLP tasks. First, I describe paired learning and inference algorithms for dynamic feature selection which accelerate inference in linear classifiers, the heart of the fastest NLP models, by 5-10 times. I then present iterated dilated convolutional neural networks (ID-CNNs), a distinct combination of network structure, parameter sharing and training procedures that increase inference speed by 14-20 times with accuracy matching bidirectional LSTMs, the most accurate models for NLP sequence labeling. Finally, I describe linguistically-informed self-attention (LISA), a neural network model that combines multi-head self-attention with multi-task learning to facilitate improved generalization to new domains. We show that incorporating linguistic structure in this way leads to substantial improvements over the previous state-of-the-art (syntax-free) neural network models for SRL, especially when evaluating out-of-domain. I conclude with a brief discussion of potential future directions stemming from my thesis work.", "source_url": "https:\/\/openalex.org\/W2982543341", "pdf_url": "", "doi": "10.7275\/15200331", "handle": "", "repository_name": "ScholarWorks@UMassAmherst (University of Massachusetts Amherst)", "source_record_id": "W2982543341", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 287, "title": "Neural Transfer Learning for Domain Adaptation in Natural Language Processing", "title_ar": "", "title_en": "", "author": "Sara Meftah", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2021, "language": "fr", "research_domain": "nlp", "research_subdomain": "transfer-learning", "categories": "machine-learning; nlp; transfer-learning", "keywords": "", "abstract": "Apprentissage par transfert neuronal pour l'adaptation aux domaines en traitement automatique de la langue Les méthodes d’apprentissage automatique qui reposent sur les Réseaux de Neurones (RNs) ont démontré des performances de prédiction qui s'approchent de plus en plus de la performance humaine dans plusieurs applications du Traitement Automatique de la Langue (TAL) qui bénéficient de la capacité des différentes architectures des RNs à généraliser à partir des régularités apprises à partir d'exemples d'apprentissage. Toutefois, ces modèles sont limités par leur dépendance aux données annotées. En effet, pour être performants, ces modèles neuronaux ont besoin de corpus annotés de taille importante. Par conséquent, uniquement les langues bien dotées peuvent bénéficier directement de l'avancée apportée par les RNs, comme par exemple les formes formelles des langues. Dans le cadre de cette thèse, nous proposons des méthodes d'apprentissage par transfert neuronal pour la construction d'outils de TAL pour les langues peu dotées en exploitant leurs similarités avec des langues bien dotées. Précisément, nous expérimentons nos approches pour le transfert à partir du domaine source des textes formels vers le domaine cible des textes informels (langue utilisée dans les réseaux sociaux). Tout au long de cette thèse nous proposons différentes contributions. Tout d'abord, nous proposons deux approches pour le transfert des connaissances encodées dans les représentations neuronales d'un modèle source, pré-entraîné sur les données annotées du domaine source, vers un modèle cible, adapté par la suite sur quelques exemples annotés du domaine cible. La première méthode transfère des représentations contextuelles pré-entraînées sur le domaine source. Tandis que la deuxième méthode utilise des poids pré-entraînés pour initialiser les paramètres du modèle cible. Ensuite, nous effectuons une série d'analyses pour repérer les limites des méthodes proposées ci-dessus. Nous constatons que, même si l'approche d'apprentissage par transfert proposée améliore les résultats du domaine cible, un transfert négatif « dissimulé » peut atténuer le gain final apporté par l'apprentissage par transfert. De plus, une analyse interprétative du modèle pré-entraîné, montre que les neurones pré-entraînés peuvent être biaisés par ce qu'ils ont appris du domaine source, et donc peuvent avoir des difficultés à apprendre des « patterns » spécifiques au domaine cible. Issu de notre analyse, nous proposons un nouveau schéma d'adaptation qui augmente le modèle cible avec des neurones normalisés, pondérés et initialisés aléatoirement qui permettent une meilleure adaptation au domaine cible tout en conservant les connaissances apprises du domaine source. Enfin, nous proposons une approche d’apprentissage par transfert qui permet de profiter des similarités entre différentes tâches, en plus des connaissances pré-apprises du domaine source.", "source_url": "https:\/\/openalex.org\/W7149540086", "pdf_url": "https:\/\/theses.hal.science\/tel-03206378\/document", "doi": "10.70675\/9111562bzfa37z4fd8zae6bz2e8ae8d5d403", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7149540086", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 292, "title": "Resources and Applications for Dialectal Arabic: the Case of Levantine", "title_ar": "", "title_en": "Resources and Applications for Dialectal Arabic: the Case of Levantine", "author": "Chatrine Qwaider ; Kathrein Abu Kwaik", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Gothenburg University", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; ai-applications; transfer-learning; self-semi-supervised; transformers; llm; sentiment-analysis; arabic-nlp; ai-industry", "keywords": "", "abstract": "This is a thesis about the computational study of Dialectal Arabic (DA). In particular, the thesis studies DA, with a special emphasis on Levantine Arabic, and develops tools and resources for the computational study of Dialectal Arabic Natural Language Processing (DANLP). It investigates the creation of fine-grained resources that can be used for a variety of computational tasks, and a number of effective models that can deal with the complexity of fine-grained dialectal data. Dialect Identification (DI), as well as Sentiment Analysis (SA) are the Natural Language Processing (NLP) tasks investigated in this thesis. In the first part (Study 1 and Study 2), I study the DI task on both coarse-grained and fine-grained levels. For this reason, I build the first annotated Levantine (SHAMI) Dialect Corpus (SDC). Furthermore, I explore the ability of n-gram language models, Machine Learning (ML) algorithms and ensemble learning techniques to classify and detect 26 Arabic varieties. In the second part, I conduct a linguistic study to measure the lexical distance between MSA and DA, and between the dialects themselves. This is done to check whether transferring knowledge from one variety to another is possible. In the third part, studies 4,5 and 6, I explore Arabic Sentiment Analysis (SA). I investigate the idea of knowledge transfer between MSA and the dialects using SA as a case study. Furthermore, I implement various models such as the pre-trained language model BERT, Deep Learning (DL), ML and feature engineering approaches to detect the sentimental polarity of DA data. I introduce two valuable resources for this task, one focusing on Levantine sentiment (Shami-Senti), and the other for DA in general (ATSAD). I exploit different ways of annotation, e.g. human, lexicon-based and automatic distant supervision annotation. The last study is about choosing the best model for DI and SA. I exploit well-known models and approaches using various kinds of DA resources. The thesis contributes to the field of DANLP in a number of ways. The introduced valuable resources can be seen as a stepping stone for a deeper investigation and understanding of issues in DANLP. They are also reliable and can be used by researchers to address different NLP tasks. The cross-dialectal linguistic studies will open up prospects for researchers to fine-tune models and transfer knowledge among Arabic varieties. A big part of the contribution lies in designing DI and SA models. I implement several ML models that use feature engineering approaches and N-gram language models to identify the dialect or detect the sentiment. For DI, I design and implement an ensemble learning model that is able to handle fine-grained dialects. Additionally, I exploit the usage of DL models on different SA dialectal datasets and achieve competitive results. For both tasks, I exploit the recent pre-trained language models and perform a comparison to choose the best model. I also implement a semi-supervised approach for automatic labelling and annotating data with the help of self-training techniques to improve the performance of the dataset. These models will help researchers dive deeper into DANLP and create practical and industrial systems.", "source_url": "https:\/\/openalex.org\/W7216829323", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Gothenburg University Publications Electronic Archive (Gothenburg University)", "source_record_id": "W7216829323", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 295, "title": "Traffic flow forecasting with deep learning", "title_ar": "", "title_en": "Traffic flow forecasting with deep learning", "author": "Per Øyvind Kanestrøm", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Norway", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; deep-learning; computer-vision; ai-applications; predictive-analytics; rnn; object-detection; ai-smart-cities", "keywords": "", "abstract": "In recent years there has been a vast increase in available data with the ad- vancement of smart cities. In the domain of Intelligent Transportation Systems (ITS) this modernisation can positively effect transportation networks, thus cut- ting down travel time, increase efficacy, and reduce environmental impact from vehicles. Norwegian Public Roads Administration (NPRA) is currently deploying a new vehicle detector system named Datainn on all public roads in Norway. Datainn sends metadata on all detected vehicles in real time. This includes information about speed, gap between vehicles, weight, and classification of vehicle type. Many machine learning approaches has been researched in literature on how to forecast traffic flow information. One such approach is that of using Artificial Neural Networks (ANNs). In this research ANN based methods have been explored. This was done by first performing a state-of-the-art Structured Literature Review (SLR) on ANN methods in literature. From the review, Stacked Sparse Autoencoder (SSAE) model was compared with recent advances of Long Short-Term Memory (LSTM) and Deep Neural Network (DNN) on four different prediction horizons. The data foundation was the new Datainn system using traffic data from a highway around Norway s capitol, Oslo. Further, the model performance was assessed with extended feature vectors including more metadata from Datainn. The results found that the LSTM model always outperformed DNN and SSAE, although in general the performance characteristics was somewhat similar. Ex- tending the feature vector with more variables had a negative effect on DNN, while resulting in better performance for Recurrent Neural Network (RNN) on long-term (60 minute) forecasting horizons. For SSAE it had a slight positive effect, but not enough get better results than RNN or DNN.", "source_url": "https:\/\/openalex.org\/W2727387027", "pdf_url": "", "doi": "", "handle": "", "repository_name": "BIBSYS Brage (BIBSYS (Norway))", "source_record_id": "W2727387027", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 304, "title": "Traffic Forecasting Applications Using Crowdsourced Traffic Reports and Deep Learning", "title_ar": "", "title_en": "Traffic Forecasting Applications Using Crowdsourced Traffic Reports and Deep Learning", "author": "Ali Alammari", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "deep-learning", "research_subdomain": "cnn", "categories": "machine-learning; deep-learning; ai-applications; predictive-analytics; cnn; rnn; ai-smart-cities", "keywords": "", "abstract": "Intelligent transportation systems (ITS) are essential tools for traffic planning, analysis, and forecasting that can utilize the huge amount of traffic data available nowadays. In this work, we aggregated detailed traffic flow sensor data, Waze reports, OpenStreetMap (OSM) features, and weather data, from California Bay Area for 6 months. Using that data, we studied three novel ITS applications using convolutional neural networks (CNNs) and recurrent neural networks (RNNs). The first experiment is an analysis of the relation between roadway shapes and accident occurrence, where results show that the speed limit and number of lanes are significant predictors for major accidents on highways. The second experiment presents a novel method for forecasting congestion severity using crowdsourced data only (Waze, OSM, and weather), without the need for traffic sensor data. The third experiment studies the improvement of traffic flow forecasting using accidents, number of lanes, weather, and time-related features, where results show significant performance improvements when the additional features where used.", "source_url": "https:\/\/openalex.org\/W4407267545", "pdf_url": "https:\/\/digital.library.unt.edu\/ark:\/67531\/metadc1703305\/m2\/1\/high_res_d\/ALAMMARI-DISSERTATION-2020.pdf", "doi": "10.12794\/metadc1703305", "handle": "", "repository_name": "University of North Texas Digital Library (University of North Texas)", "source_record_id": "W4407267545", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 307, "title": "Encrypted Web Traffic Classification Using Deep Learning", "title_ar": "", "title_en": "Encrypted Web Traffic Classification Using Deep Learning", "author": "Iman Akbari", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Waterloo", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "computer-vision", "research_subdomain": "cnn", "categories": "deep-learning; computer-vision; ai-applications; cnn; rnn; ai-smart-cities", "keywords": "", "abstract": "Traffic classification is essential in network management for operations ranging from capacity planning, performance monitoring, volumetry, and resource provisioning, to anomaly detection and security. Recently, it has become increasingly challenging with the widespread adoption of encryption in the Internet, e.g., as a de-facto in HTTP\/2 and QUIC protocols. In the current state of encrypted traffic classification using Deep Learning (DL), we identify fundamental issues in the way it is typically approached. For instance, although complex DL models with millions of parameters are being used, these models implement a relatively simple logic based on certain header fields of the TLS handshake, limiting model robustness to future versions of encrypted protocols. Furthermore, encrypted traffic is often treated as any other raw input for DL, while crucial domain-specific considerations exist that are commonly ignored. In this thesis, we design a novel feature engineering approach that generalizes well for encrypted web protocols, and develop a neural network architecture based on Stacked Long Short-Term Memory (LSTM) layers and Convolutional Neural Networks (CNN) that works very well with our feature design. We evaluate our approach on a real-world traffic dataset from a major ISP and Mobile Network Operator. We achieve an accuracy of 95% in service-level classification with less raw traffic and smaller number of parameters, out-performing a state-of-the-art method by nearly 50% fewer false classifications. We show that our DL model generalizes for different classification objectives and encrypted web protocols. We also evaluate our approach on a public QUIC dataset with finer and application-level granularity in labeling, achieving an overall accuracy of 99%.", "source_url": "https:\/\/openalex.org\/W3162766776", "pdf_url": "", "doi": "", "handle": "", "repository_name": "UWSpace (University of Waterloo)", "source_record_id": "W3162766776", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 312, "title": "Strategic Design of Smart Bike-Sharing Systems for Smart Cities", "title_ar": "", "title_en": "Strategic Design of Smart Bike-Sharing Systems for Smart Cities", "author": "Huthaifa I. Ashqar", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "machine-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; ai-applications; predictive-analytics; ai-smart-cities", "keywords": "", "abstract": "Traffic congestion has become one of the major challenging problems of modern life in many urban areas. This growing problem leads to negative environmental impacts, wasted fuel, lost productivity, and increased travel time. In big cities, trains and buses bring riders to transit stations near shopping and employment centers, but riders then need another transportation mode to reach their final destination, which is known as the last mile problem. A smart bike-sharing system (BSS) can help address this problem and encourage more people to ride public transportation, thus relieving traffic congestion. At the strategic level, we start with proposing a novel two-layer hierarchical classifier that increases the accuracy of traditional transportation mode classification algorithms. In the transportation sector, researchers can use smartphones to track and obtain information of multi-mode trips. These data can be used to recognize the user's transportation mode, which can be then utilized in several different applications; such as planning new BSS instead of using costly surveys. Next, a new method is proposed to quantify the effect of several factors such as weather conditions on the prediction of bike counts at each station. The proposed approach is promising to quantify the effect of various features on BSSs in cases of large networks with big data. Third, these resulted significant features were used to develop state-of-the-art toolbox algorithms to operate BSSs efficiently at two levels: network and station. Finally, we proposed a quality-of-service (QoS) measurement, namely Optimal Occupancy, which considers the impact of inhomogeneity in a BSS. We used one of toolbox algorithms modeled earlier to estimate the proposed QoS. Results revealed that the Optimal Occupancy is beneficial and outperforms the traditionally-known QoS measurement.", "source_url": "https:\/\/openalex.org\/W3016951819", "pdf_url": "", "doi": "", "handle": "", "repository_name": "VTechWorks (Virginia Tech)", "source_record_id": "W3016951819", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 313, "title": "Scalable and accurate forecasting for smart cities", "title_ar": "", "title_en": "Scalable and accurate forecasting for smart cities", "author": "Pasan Karunaratne", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Melbourne", "university_source": "repository_host", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "responsible-ai", "research_subdomain": "xai", "categories": "machine-learning; responsible-ai; ai-applications; predictive-analytics; xai; ai-smart-cities", "keywords": "", "abstract": "Cities are getting bigger, better and smarter. The increased connectivity of people and devices and the availability of cheap sensors has led to a surge in public and government interest in smart city initiatives. This public interest, along with the recent increased interest in machine learning techniques has led to growing research focus into the mining and analysis of data in smart city settings. Much of the analysis in smart city settings is based on forecasting on time series data recorded by smart sensors for planning purposes. For example, utility companies can use electricity load forecasting on smart meter data for capacity planning, and prediction of pedestrian counts and passenger flow in public transportation systems can help in planning to reduce traffic congestion. Though forecasting in smart city settings yields such benefits, it also entails unique challenges, such as challenges related to multi-step prediction, challenges related to low quality training data due to sensors encountering vandalism, malfunction or communication failures, and challenges in maintaining predictive throughput in systems involving increasingly larger numbers of smart sensors. Improving accuracy is a primary goal in any forecasting task, which is especially challenging in multi-step prediction scenarios. We address this challenge by providing new methods to incorporate prior knowledge uniquely relevant to smart cities, such as the periodic behaviour of sensor time series data over the Monday-Friday working week. Specifically, we propose novel kernel function compositions which can incorporate such prior knowledge to kernel-based Bayesian forecasting techniques, with the goal of improving prediction accuracy and robustness to spurious data. We develop our kernel compositions for the state of the art Gaussian Process Regression technique. The new kernel compositions we develop enable prior knowledge relating to multiple periodic effects of the working week (e.g. daily, weekly, holiday effects) and their interactions to be incorporated in the same model. We also provide methods to mitigate the effects of convergence to local optima in the optimisation process over the hyperparameters used in the Gaussian Process models. We address the challenges relating to missing training data in smart city settings by making use of data of other related sensors (which may have more complete data) to mitigate the impact the low quality data has on prediction accuracy. To this end, we develop multi-task learning methods (which are able to learn joint representations from multiple sensors) to improve Gaussian Process Regression prediction accuracy with missing training data values. We also provide equivalent expressions to our multi-task learning methods as combinations of commonly used kernel functions in Gaussian Processes. This enables the straightforward implementation of these methods in popular machine learning toolkits. We address the scalability challenge of large volumes of sensor data in two steps. One, we focus on an interpretable label-based forecasting algorithm which allows for high-throughput predictions due to the minimal number of operations needed to be done in the forecasting stage. We perform numerous enhancements on this algorithm in order to improve its prediction accuracy, including filtering, windowing and ensembling methods as well as methods of incorporating exogenous variables. Our scalable forecasting methods are then developed using this enhanced base algorithm. We develop methods which enable the initial step of the algorithm to be performed using algorithms developed for stream processing, which not only allows for the algorithm to be parallelised across multiple machines, but also enables it to run on real-time data streams. We address the scalability challenges in scenarios with both a single fast stream and a large number of streams, especially with regard to synchronisation issues between multiple machines. We demonstrate the effectiveness of our methods on multiple real-world publicly available datasets to illustrate the potential generalisability of our techniques.", "source_url": "https:\/\/openalex.org\/W2897407191", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Minerva Access (University of Melbourne)", "source_record_id": "W2897407191", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 324, "title": "Promoting Artificial Intelligence Education in K – 12 through Pre-service Teachers’ Development", "title_ar": "", "title_en": "Promoting Artificial Intelligence Education in K – 12 through Pre-service Teachers’ Development", "author": "Olumide Ojo", "supervisor": "", "degree_type": "master", "degree_evidence": "نص السجل: «master's»", "university": "University of Eastern Finland", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "responsible-ai", "research_subdomain": "fairness", "categories": "responsible-ai; ai-applications; fairness; ai-education", "keywords": "", "abstract": "This study was conducted to develop pre-service CS teachers' artificial intelligence (AI) knowledge through a professional development (PD) program.This research further examined how 45 teacher candidates learned AI concepts, their perspectives as well as their concerns and suggestions regarding implementing AI lessons in classroom.Both quantitative and qualitative methodology were employed to answer the two research questions that guided the study.A duly validated instrument was used to collect data for the quantitative part while qualitatively, semistructured interviews were conducted after the PD program in a public university in Nigeria.The analytical approach used to analyze the data retrieved includes paired sample t-test and thematic analysis.Findings revealed some influence of the PD program on the pre-service teachers' perceptions towards AI.Based on the measured variables, teachers' competence, and skills (Ability) significantly increased post PD, while teachers' cognitive readiness (cognition), which includes knowledge about how AI functions does not increase.Perceived threats about AI decreased post-training, indicating a positive impact on teachers' confidence in learning AI.This study emphasizes the necessity of AI literacy beyond short courses for boosting cognition.Concerns related to ethical issues, infrastructure challenges, and a lack of technical expertise were uncovered as significant barriers to implementing AI lessons in K-12 classrooms, prompting thoughtful suggestions from pre-service teachers.The findings from this study have implications for the development of AI education in the K-12 context and teacher education.iv Acknowledgement Embarking on this master's degree has truly been a journey.As I approach its completion, I am indebted to appreciate individuals who have played a vital role in this intellectual odyssey.I want to give glory to the Almighty God, my helper and sustainer who has been there for me through thick and thin, and for His grace has seen me throughout this program.My unflinching appreciation also goes to my able and amiable supervisor, Dr. Ismaila Sanusi whose unwavering guidance, constructive critiques, and relentless commitment to pushing the boundaries of knowledge have been the foundation of this endeavour.His guidance shaped the rigour of this thesis but also inspired my passion to pursue excellence.To Dr. Solomon Oyelere, Bar.Kola Abiri, Mr. Amuni Olayinka and Mr. Lewis Akinwunmi, I want to say a big thank you for the mentorship and concern to see that I am making progress.The resources and support provided by the University of Eastern Finland, especially the School of Computing have played a role in achieving this academic feat.The enabling academic environment and the timeless assistance from my lecturers and the also program coordinator -Oili Kohonen have paved the way for growth and innovative thinking.I would like to express sincere gratitude to my precious jewel Bukola Ojo and my lovely children Joanna and Jedidah for their prayers, inspiration and understanding.Their endless support throughout the challenging phases of my academic pursuit has been my anchor.Their confidence in my strength has served as a propelling force and a constant reminder of the importance of perseverance.To everyone who has supported me in this course, I extend my thanks for the fellowship that has helped in the actualisation of this academic goal.Your support has been of immense benefit.In summary, have concluded that in unity, we can achieve a lot.I am really grateful to each person mentioned and those I could not mention for their endless support throughout this journey.", "source_url": "https:\/\/openalex.org\/W7215644623", "pdf_url": "https:\/\/erepo.uef.fi\/bitstreams\/38c3bac6-e930-4fa4-a300-17eb97d4dcc7\/download", "doi": "", "handle": "", "repository_name": "UEF eRepo (University of Eastern Finland)", "source_record_id": "W7215644623", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 334, "title": "Predictive Scheduling of Incoming Goods Using Machine Learning and Deep Learning Methods", "title_ar": "", "title_en": "", "author": "Johannes Simon Michael Rechenauer", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "TU Wien", "university_source": "repository_host", "college": "", "department": "", "country": "Austria", "governorate": "", "publication_year": 2026, "language": "de", "research_domain": "deep-learning", "research_subdomain": "data-mining", "categories": "machine-learning; deep-learning; ai-applications; data-mining; predictive-analytics; ai-education", "keywords": "", "abstract": "Die vorliegende Arbeit untersucht den Einsatz von Machine-Learning- und Deep-Learning-Verfahren zur prädiktiven Terminierung von Wareneingängen. Ausgangspunkt bilden betriebliche Herausforderungen, welche durch Lieferabweichungen entstehen und eine zuverlässige Materialversorgung erschweren. Die zentrale Zielsetzung besteht somit darin, Wareneingänge mithilfe datengetriebener Prognosemodelle so zu terminieren, dass Materialbedarfe gedeckt und gleichzeitig Lagerbestände möglichst gering gehalten werden. Methodisch orientiert sich diese Arbeit am CRISP-DM-Prozess, welcher als Rahmen dient, um die betriebliche Problemstellung systematisch in ein Data-Mining-Problem zu überführen. Dafür wird die Problemstellung zunächst konzeptionell in einem Unternehmen verortet. Gleichzeitig werden Grundlagen in den Bereichen logistische Planung, Data Mining, Deep Learning und Machine Learning gelegt. Darauf aufbauend wird ein Data-Mining-Projekt in einem realen Unternehmensumfeld ausgearbeitet. Dabei wird die Wareneingangsterminierung als betriebliche Problemstellung eingeordnet und als Klassifikationsproblem formuliert. Im Rahmen der Modellierungsphase werden daraufhin verschiedene Boosting-Modelle und Deep-Learning-Ansätze verglichen. Der Fokus liegt in diesem Zusammenhang nicht auf der isolierten Leistung einzelner Algorithmen, sondern auf der Leistungsfähigkeit der gesamten Modell-Pipeline. Diese umfasst die modellspezifische Datenvorverarbeitung, das Modelltraining, die Hyperparameter-Optimierung sowie die anschließende Evaluation. Dadurch werden die Verfahren unter möglichst praxisnahen Bedingungen hinsichtlich ihrer Vorhersageleistung und praktischen Anwendbarkeit miteinander verglichen. Die Ergebnisse zeigen, dass Boosting-Verfahren im untersuchten Anwendungsfall am besten abschneiden. Insbesondere die auf LightGBM basierende Pipeline erzielt die beste Kombination aus Vorhersageleistung und Effizienz, während die getesteten Deep-Learning-Modelle trotz längerer Trainingszeiten schwächere Ergebnisse liefern. Aufbauend auf den Ergebnissen des Modellvergleichs wird dargestellt, wie prognostizierte Verzögerungsklassen in konkrete Vorschläge zur Wareneingangsterminierung überführt und im Rahmen des CRISP-DM-Prozesses in einem Unternehmen implementiert werden können. Somit leistet die Arbeit einerseits einen Beitrag zur praxisnahen Ausarbeitung eines Data-Mining-Projekts und verdeutlicht andererseits, dass moderne Boosting-Verfahren insbesondere bei kleinen tabellarischen Datensätzen Deep-Learning-Architekturen überlegen sein können.", "source_url": "https:\/\/openalex.org\/W7223559876", "pdf_url": "https:\/\/repositum.tuwien.at\/bitstream\/20.500.12708\/230957\/1\/Rechenauer%20Johannes%20Simon%20Michael%20-%202026%20-%20Praediktive%20Terminierung%20von...pdf", "doi": "10.34726\/hss.2026.144705", "handle": "", "repository_name": "reposiTUm (TU Wien)", "source_record_id": "W7223559876", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 342, "title": "Exploration of Deep Learning Models on Streamflow Simulations", "title_ar": "", "title_en": "Exploration of Deep Learning Models on Streamflow Simulations", "author": "Jinyang Li", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "computer-vision", "research_subdomain": "image-segmentation", "categories": "machine-learning; deep-learning; computer-vision; ai-applications; predictive-analytics; rnn; image-segmentation; ai-agriculture", "keywords": "", "abstract": "Streamflow simulation and forecasting is an important approach for water resources management and flood mitigation, and both physically-based and data-driven models have been used to predict the streamflow at daily and hourly scale. Although physically-based models can help understand the underlying mechanisms of the hydrological process based on mathematical and physical equations, the performance of the models largely depends on the availability and quality of the spatial information. The computation cost also limits its application in the fast, reliable and accurate streamflow simulation and forecasting. In contrast, data-driven models, especially deep learning models, are capable of learning the nonlinear relationships between the inputs and outputs without explicitly referring to the physical process. In addition, deep learning techniques, powered by graphics processing units (GPUs) can allow researchers to train models 10 or 20 times faster, which is more promising for inputs-outputs simulation. Currently, the deep learning models rely heavily on prior streamflow information to make predictions, while the hydrological observations are limited in the real world applications. In this Thesis, the effectiveness of deep learning models on streamflow simulation without using the previous streamflow information was explored. In specific, the three deep learning models the ANN, RNN and LSTM models were used to simulate the streamflow at multiple lead time and different sites using the precipitation and soil moisture. In addition, the effectiveness of deep learning models on learning from a physically-based model to generate pixel-to-pixel streamflow mapping has also been tested.All three deep learning models can achieve reliable and accurate streamflow simulations at the lead time of 1-5 h at the Ninnekah site with Pearson correlation coefficient (r) over 0.9 for all three models. Based on the prediction of the peak and low flow on different sites, the LSTM model shows more consistent and reliable streamflow simulation results and can work better in different basin scales. For the U-net model, it can learn the patterns of ParFlow-CLM model well and achieves average r of 0.770 and 0.820 in the whole domain for validation and calibration, respectively. However, the models have limitations in predicting the peak due to the feature of the imbalanced data.", "source_url": "https:\/\/openalex.org\/W3188379214", "pdf_url": "", "doi": "", "handle": "", "repository_name": "eScholarship (California Digital Library)", "source_record_id": "W3188379214", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 343, "title": "ON META-NETWORKS, DEEP LEARNING, TIME AND JIHADISM", "title_ar": "", "title_en": "", "author": "Gian Maria Campedelli", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2020, "language": "it", "research_domain": "deep-learning", "research_subdomain": "", "categories": "deep-learning", "keywords": "", "abstract": "Il terrorismo di stampo jihadista rappresenta una minaccia per la società e una sfida per gli scienziati interessati a comprenderne la complessità. Questa complessità richiede costantemente nuovi sviluppi in termini di ricerca sul terrorismo. Migliorare la conoscenza empirica rispetto a tale fenomeno può potenzialmente contribuire a sviluppare applicazioni concrete e, in ultima istanza, a prevenire danni all’uomo. In considerazione di tali aspetti, questa tesi presenta un nuovo quadro metodologico che integra scienza delle reti, modelli stocastici e apprendimento profondo per far luce sul terrorismo jihadista sia a livello esplicativo che predittivo. In particolare, questo lavoro compara e analizza le organizzazioni jihadiste più attive a livello mondiale (ovvero lo Stato Islamico, i Talebani, Al Qaeda, Boko Haram e Al Shabaab) per studiarne i pattern comportamentali e predirne le future azioni. Attraverso un impianto teorico che si poggia sulla concentrazione spaziale del crimine e sulle prospettive strategiche del comportamento terroristico, questa tesi persegue tre obiettivi collegati utilizzando altrettante tecniche ibride. In primo luogo, verrà esplorata la complessità operativa delle organizzazioni jihadiste attraverso l’analisi di matrici stocastiche di transizione e verrà presentato un nuovo coefficiente, denominato “Normalized Transition Similarity”, che misura la somiglianza fra paia di gruppi in termini di dinamiche operative. In secondo luogo, i processi stocastici di Hawkes aiuteranno a testare la presenza di meccanismi di dipendenza temporale all’interno delle più comuni sotto-sequenze strategiche di ciascun gruppo. Infine, il framework integrerà la meta-reti complesse e l’apprendimento profondo per classificare e prevedere i target a maggiore rischio di essere colpiti dalle organizzazioni jihadiste durante i loro futuri attacchi. Per quanto riguarda i risultati, le matrici stocastiche di transizione mostrano che i gruppi terroristici possiedono un ricco e complesso repertorio di combinazioni in termini di armi e obiettivi. Inoltre, i processi di Hawkes indicano la presenza di diffusa self-excitability nelle sequenze di eventi. Infine, i modelli predittivi che sfruttano la flessibilità delle serie temporali derivanti da grafi dinamici e le reti neurali Long Short-Term Memory forniscono risultati promettenti rispetto ai target più a rischio. Nel complesso, questo lavoro ambisce a dimostrare come connessioni astratte e nascoste fra eventi possano essere fondamentali nel rivelare le meccaniche del comportamento jihadista e come processi memory-like (ovvero molteplici comportamenti ricorrenti, interconnessi e non randomici) possano risultare estremamente utili nel comprendere le modalità attraverso cui tali organizzazioni operano.", "source_url": "https:\/\/openalex.org\/W3013089364", "pdf_url": "", "doi": "10.83049\/unicatt\/publicatt\/10280_70552", "handle": "", "repository_name": "", "source_record_id": "W3013089364", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 348, "title": "Evaluation and validation of deep learning strategies for bioimage analyses", "title_ar": "", "title_en": "Evaluation and validation of deep learning strategies for bioimage analyses", "author": "Dennis Segebarth", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Würzburg University", "university_source": "repository_host", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "computer-vision", "research_subdomain": "fairness", "categories": "deep-learning; computer-vision; responsible-ai; fairness", "keywords": "", "abstract": "Significant advances in fluorescence imaging techniques enable life scientists today to gain insights into biological systems at an unprecedented scale. The interpretation of image features in such bioimage datasets and their subsequent quantitative analysis is referred to as bioimage analysis. A substantial proportion of bioimage analyses is still performed manually by a human expert - a tedious process that is long known to be subjective. Particularly in tasks that require the annotation of image features with a low signal-to-noise ratio, like in fluorescence images of tissue samples, the inter-rater agreement drops. However, like any other scientific analysis, also bioimage analysis has to meet the general quality criteria of quantitative research, which are objectivity, reliability, and validity. Thus, the automation of bioimage analysis with computer-aided approaches is highly desirable. Albeit conventional hard-coded algorithms are fully unbiased, a human user has to set its respective feature extraction parameters. Thus, also these approaches can be considered subjective. Recently, deep learning (DL) has enabled impressive advances in computer vision research. The predominant difference between DL and conventional algorithms is the capability of DL models to learn the respective task on base of an annotated training dataset, instead of following user-defined rules for feature extraction. This thesis hypothesized that DL can be used to increase the objectivity, reliability, and validity of bioimage analyses, thus going beyond mere automation. However, in absence of ground truth annotations, DL models have to be trained on manual and thus subjective annotations, which could cause the model to incorporate such a bias. Moreover, model training is stochastic and even training on the same data could result in models with divergent outputs. Consequently, both the training on subjective annotations and the model-to-model variability could impair the quality of DL-based bioimage analyses. This thesis systematically assessed the impacts of these two limitations experimentally by analyzing fluorescence signals of a protein called cFOS in mouse brain sections. Since the abundance of cFOS correlates with mouse behavior, behavioral analyses could be used for cross-validation of the bioimage analysis results. Furthermore, this thesis showed that pooling the input of multiple human experts during model training and integration of multiple trained models in a model ensemble can mitigate the impact of these limitations. In summary, the present study establishes guidelines for how DL can be used to increase the general quality of bioimage analyses.", "source_url": "https:\/\/openalex.org\/W3199015547", "pdf_url": "https:\/\/nbn-resolving.org\/urn:nbn:de:bvb:20-opus-243728", "doi": "10.25972\/opus-24372", "handle": "", "repository_name": "Online Publication Service of Würzburg University (Würzburg University)", "source_record_id": "W3199015547", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 365, "title": "Application of machine learning in production scheduling", "title_ar": "", "title_en": "Application of machine learning in production scheduling", "author": "Georg Faustmann", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "TU Wien", "university_source": "repository_host", "college": "", "department": "", "country": "Austria", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "machine-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; ai-applications; predictive-analytics; ai-industry", "keywords": "", "abstract": "For human experts, it is often too hard or too time-consuming to manually detect patterns in big data sets. Machine learning is applied in many areas to detect such patterns. Its applications are by no means limited to research, as machine learning also plays a big role in the industrial sector. This thesis applies machine learning to the following two topics. The first part of the thesis deals with product quality classification for automotive paint shops. The second part investigates automated parameter configuration for dispatching rules that are used in machine scheduling. We use a binary classification to predict the product quality of an automotive paint shop based on its scheduling data. We propose a set of features to characterize the production process. These features are used to classify whether or not the quality of the product is satisfactory. We can show that the best model we found performs better than a baseline model on an unseen data set. In the automated parameter configuration part of the thesis, we investigate machine learning methods based on multi-target regression to automatically configure dispatching rules for real-life planning scenarios where multiple objectives are considered. We propose a novel set of features to characterize instances of the parallel machine scheduling problem, and describe how supervised learning can be used to obtain optimized parameter configurations for given machine scheduling instances. Experimental results show that our approach can obtain high-quality solutions for real-life scheduling scenarios in short run times.", "source_url": "https:\/\/openalex.org\/W3183329507", "pdf_url": "https:\/\/repositum.tuwien.at\/bitstream\/20.500.12708\/6541\/2\/Faustmann%20Georg%20-%202019%20-%20Application%20of%20machine%20learning%20in%20production...pdf", "doi": "10.34726\/hss.2019.68471", "handle": "", "repository_name": "reposiTUm (TU Wien)", "source_record_id": "W3183329507", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 368, "title": "Applications of Machine Learning in Mental Healthcare", "title_ar": "", "title_en": "Applications of Machine Learning in Mental Healthcare", "author": "Elena Davcheva", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "medical-ai", "research_subdomain": "", "categories": "machine-learning; medical-ai", "keywords": "", "abstract": "This thesis summarizes three studies in the area of machine learning applications within mental heathcare, specifically in the area of treatments and diagnostics. Mental healthcare today is challenging to provide worldwide because of a stark rise in demand for services. Traditional healthcare structures cannot keep up with the demand and information systems have the potential to fill in this gap. The thesis explores online mental health forums as a digital mental health platform and the possibility to automate treatments and diagnostics based on user-shared information.", "source_url": "https:\/\/openalex.org\/W3161669028", "pdf_url": "https:\/\/tuprints.ulb.tu-darmstadt.de\/17879\/1\/Applications%20of%20Machine%20Learning%20in%20Mental%20Healthcare_Davcheva.pdf", "doi": "10.26083\/tuprints-00017879", "handle": "", "repository_name": "Technischen Universität Darmstadt", "source_record_id": "W3161669028", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 369, "title": "Machine Learning for Materials Science", "title_ar": "", "title_en": "Machine Learning for Materials Science", "author": "Bertrand Rouet‐Leduc", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Cambridge", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "machine-learning", "research_subdomain": "quantum-ml", "categories": "machine-learning; ai-applications; quantum-ml; predictive-analytics; ai-industry", "keywords": "", "abstract": "Machine learning is a branch of artificial intelligence that uses data to automatically build inferences and models designed to generalise and make predictions. In this thesis, the use of machine learning in materials science is explored, for two different problems: the optimisation of gallium nitride optoelectronic devices, and the prediction of material failure in the setting of laboratory earthquakes. Light emitting diodes based on III-nitrides quantum wells have become ubiquitous as a light source, owing to their direct band-gap that covers UV, visible and infra-red light, and their very high quantum efficiency. This efficiency originates from most electronic transitions across the band-gap leading to the emission of a photon. At high currents however this efficiency sharply drops. In chapters 3 and 4 simulations are shown to provide an explanation for experimental results, shedding a new light on this drop of efficiency. Chapter 3 provides a simple and yet accurate model that explains the experimentally observed beneficial effect that silicon doping has on light emitting diodes. Chapter 4 provides a model for the experimentally observed detrimental effect that certain V-shaped defects have on light emitting diodes. These results pave the way for the association of simulations to detailed multi-microscopy. In the following chapters 5 to 7, it is shown that machine learning can leverage the use of device simulations, by replacing in a targeted and efficient way the very labour intensive tasks of making sure the numerical parameters of the simulations lead to convergence, and that the physical parameters reproduce experimental results. It is then shown that machine learning coupled with simulations can find optimal light emitting diodes structures, that have a greatly enhanced theoretical efficiency. These results demonstrate the power of machine learning for leveraging and automatising the exploration of device structures in simulations. Material failure is a very broad problem encountered in a variety of fields, ranging from engineering to Earth sciences. The phenomenon stems from complex and multi-scale physics, and failure experiments can provide a wealth of data that can be exploited by machine learning. In chapter 8 it is shown that by recording the acoustic waves emitted during the failure of a laboratory fault, an accurate predictive model can be built. The machine learning algorithm that is used retains the link with the physics of the experiment, and a new signal is thus discovered in the sound emitted by the fault. This new signal announces an upcoming laboratory earthquake, and is a signature of the stress state of the material. These results show that machine learning can help discover new signals in experiments where the amount of data is very large, and demonstrate a new method for the prediction of material failure.", "source_url": "https:\/\/openalex.org\/W2765454343", "pdf_url": "", "doi": "10.17863\/cam.13918", "handle": "", "repository_name": "Apollo (University of Cambridge)", "source_record_id": "W2765454343", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 373, "title": "Machine learning in neonatal intensive care", "title_ar": "", "title_en": "Machine learning in neonatal intensive care", "author": "Olli‐Pekka Rinta‐Koski", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Aalto University", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "medical-ai", "research_subdomain": "fairness", "categories": "machine-learning; medical-ai; responsible-ai; predictive-analytics; fairness", "keywords": "", "abstract": "This thesis deals with the applications of machine learning in the context of neonatal intensive care. The main focus is on prediction using multichannel time series data. The purpose of developing prediction methods is to create data-driven tools for helping care personnel to make more informed decisions. Preterm infants, born before 37 weeks of gestation, are subject to many developmental issues and health problems. Very Low Birth Weight infants, with a birth weight under 1500 g, are the most afflicted in this group. These infants require treatment in the neonatal intensive care unit before they are mature enough for hospital discharge. The neonatal intensive care unit is a data-intensive environment, where multi-channel physiological data is gathered from patients using a number of sensors to construct a comprehensive picture of the patients' vital signs. Neonatal intensive care requires combining information from multiple sources, often with severe and far-reaching consequences for the patient. Real-time data is available on multiple monitor screens in the ward, and is used by care personnel to help them make informed decisions. Analysis of trends and vast data sets, possibly spanning thousands of patients, requires the use of machine learning methods. In the research done for this thesis we have applied machine learning methods, in particular Gaussian process classification, to predict neonatal in-hospital mortality and morbidities. We have used time series data collected from Very Low Birth Weight infants treated in the neonatal intensive care unit of Helsinki University Hospital between 1999 and 2013. Our results show that machine learning models based on time series data alone have predictive power comparable with standard medical scores, and combining the two results in improved predictive ability. We have also studied the effect of observer bias on recording vital sign measurements in the neonatal intensive care unit, as well as conducted a retrospective cohort study on trends in the growth of Extremely Low Birth Weight (birth weight under 1000 g) infants during intensive care.", "source_url": "https:\/\/openalex.org\/W2894079146", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Aaltodoc (Aalto University)", "source_record_id": "W2894079146", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 375, "title": "Leveraging Self-Attention in Transformer Models for Improved Retail Demand Forecasting", "title_ar": "", "title_en": "Leveraging Self-Attention in Transformer Models for Improved Retail Demand Forecasting", "author": "Rohith Sardena", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "National College of Ireland", "university_source": "repository_host", "college": "", "department": "", "country": "Ireland", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; predictive-analytics; rnn; transformers", "keywords": "", "abstract": "Inventory management and operation planning in retail sector involve accurate demand forecasting as a key element of the process. Linear models and tree-based models are typical examples of traditional forecasting models that may not work well in capturing complex, non-linear, and time-evolving demands that retail demand exhibits especially when it is affected by promotions, seasonal trends, and other external market factors. The study examines the possibility of deep learning models based on Transformers, which will use self-attention mechanisms, to improve the accuracy and scalability of retail demand forecasting systems. Model development was done using a detailed retail dataset with historical sales, inventory, pricing, discounts, promotions, weather conditions, and other contextual characteristics. A number of models were applied and tested in the study: Linear Regression, Decision Tree, Random Forest, Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Transformer. Mean Absolute Error (MAE), Root Mean Squared Error (RMSE) and Median Absolute Error (MedianAE) were used to evaluate performance. The classical models were moderate in their performance, and Linear Regression and Random Forest injected MAEs of 7.47 and 7.55 respectively, and RMSEs of more than 8.6. Deep learning models, on the contrary, showed a dramatic increase. Transformer model performed the best and its MAE was 0.81, RMSE was 0.99, and MedianAE was 0.74, which are better than LSTM (MAE: 0.83) and GRU (MAE: 0.83). The results verify that self-attention mechanism of Transformer is capable of learning long-range dependencies and complex temporal interactions, giving significant improvement on the baseline models. This study shows the ability of Transformers to be an efficient and scalable method of real-time retail demand forecasting. Further development will consider multi-step forecasting, probabilistic additions and deployment in other retail areas.", "source_url": "https:\/\/openalex.org\/W7204618203", "pdf_url": "https:\/\/norma.ncirl.ie\/9664\/2\/rohithsardenaconfigurationmanual.pdf", "doi": "", "handle": "", "repository_name": "TRAP@NCI (National College of Ireland)", "source_record_id": "W7204618203", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 376, "title": "Understanding Human-Model Alignment in Visual Social Attention with Video Transformers", "title_ar": "", "title_en": "Understanding Human-Model Alignment in Visual Social Attention with Video Transformers", "author": "Sabrina M. Lopez", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; responsible-ai; predictive-analytics; cnn; transformers; video-analysis; fairness", "keywords": "", "abstract": "Artificial Neural Networks (ANNs) are among the leading computational frameworks for approximating human visual processing across a range of visual tasks. Despite this, a notable gap remains between human and ANN performance in tasks that require social perception and cognition such as intent prediction and social interaction recognition. Among ANN architectures, vision transformers have recently emerged as particularly promising approximations of human visual processing as they exhibit a stronger shape bias and produce errors more consistent with those of human observers, and develop more globally uniform internal representations. This paper examines how transformers align with human perception when identifying social interactions in videos, in which two agents help, hinder, or have neutral interactions with each other in a 2D simulated environment. With the use of a dataset containing videos paired with human gaze recordings captured during the classification of helping and hindering social interactions, we assess the degree of human-model alignment through both classification accuracy and visual attention metrics across different models (i.e., TimeSformer, ViViT, V-JEPA 2), different pre-training datasets (e.g., Something Something V2, Kinetics 400), and different visualization methods (i.e., different GradCam methods and integrated gradients). Our findings suggest TimeSformer and ViViT models have strong potential as cognitive models of social interaction, with attention patterns — both static and dynamic — that align meaningfully with those of human participants. Meanwhile, despite their higher classification accuracy, V-JEPA 2 models achieve saliency alignment at or below that of convolutional neural network baselines. Notably, classification performance is not a reliable predictor of human-model attention alignment. Regarding pretraining datasets, larger and more diverse datasets tend to improve saliency alignment for classification-supervised models, though the optimal visualization method for extracting alignment varies by architecture.", "source_url": "https:\/\/openalex.org\/W7155351216", "pdf_url": "", "doi": "10.18130\/08r1-yn85", "handle": "", "repository_name": "Libra", "source_record_id": "W7155351216", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 378, "title": "Counter-Rotating Aperture Resolution: A Formal Stream-Processing Model for SHA-256 Recurrence, Root-Cover Coupling, and Transformer Language Attention", "title_ar": "", "title_en": "Counter-Rotating Aperture Resolution: A Formal Stream-Processing Model for SHA-256 Recurrence, Root-Cover Coupling, and Transformer Language Attention", "author": "Dean Kulik", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; nlp; predictive-analytics; transformers; llm", "keywords": "", "abstract": "Counter-Rotating Aperture Resolution: A Formal Stream-Processing Model for SHA-256 Recurrence, Root-Cover Coupling, and Transformer Language Attention Driven By Dean A. Kulik May 2026 Abstract This paper formalizes Counter-Rotating Aperture Resolution (CRAR): a model in which a fixed aperture and a moving data surface produce output through relative phase contact. The central claim is operational rather than metaphoric: a static output, or “still,” cannot identify the stream that generated it. A stream is an operator, recurrence, or transition law, and it is exposed only through controlled perturbation across the aperture. The theory is developed in three stages. First, we define stills, streams, apertures, and counter-rotating domains using a train-and-circle model. A processor may remain externally fixed while an oppositely moving surface passes beneath it; computation occurs at the contact point, not in absolute translation. Second, we instantiate the model in the SHA-256 message schedule. The lag set and seed width generate the aperture complement which organizes the direct-entry taxonomy. The first two expanded words have parent apertures whose unique intersection is . Thus is the Root-Cover Axis: the unique seed word that directly seeds both expansion roots and , has , has clean echo set , and saturates immediately at . The result is then correctly bounded: is the first triple-entry, length-sensitive face-selection hinge, while is the global coupling axis. Third, we translate the mathematics into transformer attention. Causal language models process symbolic streams by passing prompt tokens across fixed weights and a causal mask. Attention is not simple forward reading; it is counter-rotating aperture contact between a forward token stream and a backward-facing context field. We distinguish a forbidden seam, implemented by the causal mask for , from a balance seam, such as equal-score or orthogonality surfaces in . The SHA-to-transformer mapping is stated as a testable transport hypothesis, not as a completed proof. The result is a unified mathematical paper on why the word is the still, meaning is the stream crossing it, and controlled perturbation is the method that reveals the operator. 1. Scope and Thesis The working principle is: A still is a projected state. A stream is the transition law that moves states. If only the final state is observed, the underlying operator is generally non-identifiable. This matters for cryptographic schedules, language models, and any symbolic processor whose visible output hides the internal recurrence. In SHA-256, the final digest is a still. The message schedule is the stream. In language, a word token is a still. Meaning is the stream crossing the token through context, memory, and prediction. In both cases, the correct object is not the output value, but the preserved transition structure exposed by perturbation. This paper therefore studies the operator geometry rather than the final output. This paper does not claim a break of SHA-256. It does not claim that transformer attention has been fully proven equivalent to SHA-256 recurrence. It extracts the shared mathematics of aperture, counterflow, recurrence, masking, contamination, saturation, and projection. 2. Stills, Streams, and Operators Let denote a system state at discrete time . A still is a projection of a state: A stream is the transition law: or, more generally, The differential view is The still is residue. The stream is recurrence. 2.1 Static Non-Identifiability Lemma Lemma 2.1. Let be a many-to-one projection. Given only , the transition operator that generated is not uniquely identifiable. Proof. Since is many-to-one, there exist distinct states such that . There may exist distinct transition operators and and initial states satisfying Both generate the same projected still . Therefore alone cannot determine the stream operator. 2.2 Perturbation Principle A stream is exposed by controlled perturbation: Define a perturbation-response map for differentiable systems, or a Boolean contamination recurrence for discrete systems. The operator becomes visible through where motion can and cannot propagate. Thus: 3. The Train-Circle Model of Counter-Rotating Recursion Consider a train moving linearly: On the train is a circular recursive domain of radius with angular phase A point on the circle has global position The train moves forward in global time, but the circular domain returns to previous local phase states: Thus the system does not reverse external time; it returns to a previous domain state. 3.1 Radius as Phase Curvature The tangential velocity on the circle is At the center, , the point follows the train: Near the rim, phase dominates. Therefore: 3.2 Counter-Rotating Surface Now suppose an observer attempts to walk one way while the circular surface moves the opposite way. Let Then the observer has no external displacement: But the surface still passes beneath the contact point.", "source_url": "https:\/\/openalex.org\/W7162682989", "pdf_url": "", "doi": "10.5281\/zenodo.20426172", "handle": "", "repository_name": "Zenodo (CERN European Organization for Nuclear Research)", "source_record_id": "W7162682989", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 381, "title": "CONTRASTIVE ATTENTION TRANSFORMER FOR CLASSIFICATION", "title_ar": "", "title_en": "CONTRASTIVE ATTENTION TRANSFORMER FOR CLASSIFICATION", "author": "Yunjiao Bai", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Rutgers University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; self-semi-supervised; predictive-analytics; transformers", "keywords": "", "abstract": "Most traditional classification methods process each data point in isolation, ignoring the structural information embedded in the relationships among data points. This dissertation develops the contrastive attention transformer, a novel classification architecture that attends a given point separately to each candidate class, producing one representation per class, and predicts the label by contrasting these representations through a shared classifier. We develop the framework for both binary and multiclass classification, proposing three loss functions under two complementary approaches. The comparative approach enforces the relative ordering of confidences across representations through the probability difference and proportional losses. The aggregation approach maximizes joint correctness through the conditional probability loss, which we show to coincide with a forecast aggregation model from decision analysis and to be equivalent to classification on the sum of the representations, motivating a flexible concatenation architecture. A key-value caching scheme further reduces the computational cost of attending to each class separately by 50 to 60 percent for large numbers of classes. We evaluate the framework on five binary classification tasks, four multiclass classification tasks, and a controlled CIFAR-100 experiment, spanning image, text, and tabular data. The best contrastive variant outperforms a feedforward network baseline and a standard transformer baseline with unrestricted attention on every binary and multiclass task. The CIFAR-100 experiment identifies the distinguishability of the classes, rather than the number of classes, as the primary driver of the contrastive method's advantage over unrestricted attention.", "source_url": "https:\/\/openalex.org\/W7220982118", "pdf_url": "", "doi": "10.7282\/t3-ger4-v617", "handle": "", "repository_name": "Rutgers University Community Repository (Rutgers University)", "source_record_id": "W7220982118", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 383, "title": "Transformers and other Attention-Based Algorithms for Biomedicine", "title_ar": "", "title_en": "Transformers and other Attention-Based Algorithms for Biomedicine", "author": "Marco Hernández Pérez", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Salamanca", "university_source": "repository_host", "college": "", "department": "", "country": "Spain", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; self-semi-supervised; predictive-analytics; rnn; transformers; image-segmentation; medical-imaging", "keywords": "", "abstract": "[EN] This research explores the application of Transformer-based models and Hybrid Attention Mechanisms across three critical biomedical tasks: meningioma segmentation, epileptic seizure detection, and pathogenicity prediction of genomic variants. For meningioma segmentation, this research investigates refining skip connections in a U-Net architecture by incorporating Swin Transformers for Magnetic Resonance Imaging. The proposed SwinC U-Net model, evaluated on the Brain Tumor Segmentation Meningioma 2023 challenge dataset, achieved a Dice score of 0.8933±0.0016, precision of 0.9020±0.0041, and recall of 0.8939±0.0044, surpassing U-Net and Attention U-Net in precision and accuracy, while slightly underperforming in recall compared to U-Net. In epileptic seizure detection, this work addresses the challenge of real-time monitoring using microelectrode array data from rat brain slices with 4-aminopyridine-induced epileptic activity. Lightweight neural models were trained to detect seizures. The best performing model, GRU+Attention, achieved an event-level F1 score of 0.839, Jaccard index of 0.722, recall of 0.813, and precision of 0.867 on a held-out test set. The model was quantized to INT8, maintaining high event-level detection performance (F1 = 0.812, Jaccard = 0.709) and demonstrating real-time feasibility on embedded platforms like the Raspberry Pi 5 (latency: 2.61 ms) and Coral Dev Board (latency: 23.7 ms), with memory usage under 614 MB. For pathogenicity prediction, a Feature Tokenizer Transformer model was developed to classify genetic variants. The model used a set of inputs derived from next-generation sequencing data from the CLINVAR database, after being processed through the regular pipeline consisting of quality control, alignment, variant calling, filtering, indexing, and annotation. The semi-supervised approach allowed the model to extract insight from uncertain-labeled data. The model achieved a precision of 0.98, sensitivity of 0.92, specificity of 0.99, and F1 score of 0.95, outperforming ClinPred and REVEL in precision and specificity, but with slightly lower sensitivity than ClinPred. These studies demonstrate the versatility and efficacy of Attention-based models in handling diverse biomedical signal modalities. The results underscore the potential of Transformer architectures to advance biomedical data analysis, enhancing diagnostics and personalized medicine.", "source_url": "https:\/\/openalex.org\/W4414483146", "pdf_url": "", "doi": "10.14201\/gredos.167164", "handle": "", "repository_name": "Gredos (University of Salamanca)", "source_record_id": "W4414483146", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 387, "title": "Explainable Transformers in Natural Language Processing", "title_ar": "", "title_en": "Explainable Transformers in Natural Language Processing", "author": "George Mihaila", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "deep-learning; nlp; medical-ai; responsible-ai; transformers; sentiment-analysis; speech-recognition; xai; fairness", "keywords": "", "abstract": "Transformer-based models have revolutionized natural language processing while becoming increasingly opaque in their decision-making processes. As these systems deploy in high-stakes domains such as healthcare, legal decision-making, and content moderation, their lack of transparency poses critical challenges, particularly where explainability is legally mandated or ethically required. This dissertation introduces ExpNet, a novel framework that learns to generate human-interpretable explanations from transformer attention mechanisms through supervised learning. Unlike existing approaches that apply fixed mathematical heuristics or rely on computationally expensive perturbation methods, ExpNet treats explanation generation as a learning problem. The framework extracts attention features and processes them through a lightweight neural network trained on human-annotated rationales, discovering optimal mappings from attention patterns to token importance scores. ExpNet demonstrates strong cross-task generalization capability through rigorous evaluation on diverse natural language processing tasks including sentiment analysis, grammatical acceptability judgment, and hate speech detection. When trained on tasks from certain domains and evaluated on entirely different tasks, ExpNet consistently outperforms thirteen baseline methods spanning model-agnostic, gradient-based, and attention-based approaches. This generalization achieved without task-specific retraining suggests that transformer attention encodes universal linguistic patterns, enabling the framework to explain models across a broad range of applications beyond its training domains. Quantitative results show ExpNet achieves substantially higher alignment with human rationales than all baselines across varied linguistic phenomena. The framework generates explanations orders of magnitude faster than perturbation-based methods, making it suitable for real-time deployment. This combination of superior explanation quality and computational efficiency addresses the critical gap between interpretability research and practical production requirements. The dissertation develops extensive analysis of explanation evaluation challenges in natural language processing. By learning from human reasoning, generalizing across diverse tasks, and maintaining computational efficiency, ExpNet provides a pathway toward transparent transformer-based systems suitable for high-stakes applications where explainability is imperative.", "source_url": "https:\/\/openalex.org\/W7166919207", "pdf_url": "", "doi": "", "handle": "", "repository_name": "University of North Texas Digital Library (University of North Texas)", "source_record_id": "W7166919207", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 390, "title": "Towards the Renormalization of Transformer Language Models", "title_ar": "", "title_en": "Towards the Renormalization of Transformer Language Models", "author": "Jingcheng Niu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Toronto", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "deep-learning; nlp; transformers; llm", "keywords": "", "abstract": "Transformer language models (LMs), the cornerstone of the popular large language models (LLMs), have revolutionised the fields of artificial intelligence (AI) and natural language processing (NLP). However, we still understand relatively little about their computational mechanisms and the reasons for their effectiveness. The internal workings of these models remain enigmatic \"black boxes.\" In this thesis, I present a framework for interpreting LMs, which can be aptly described using the term \"renormalization.\" Inherent model components — such as layers, attention heads, MLPs, or neurons — should serve as the base units of abstraction for the renormalization process. This approach is preferable to using familiar linguistic or cognitive structures, as there is insufficient evidence to suggest that the internal mechanisms of LMs operate in a manner analogous to human linguistic or cognitive frameworks. Furthermore, our analysis of the Knowledge Neuron (KN) thesis suggests that neurons exhibit a remarkable degree of idiosyncrasy in the types of function and information they associate with. This finding underscores the necessity of conducting the coarse-graining process at the granularity of individual neurons. However, our examination also reveals critical limitations of the KN thesis, suggesting that simply decomposing the model without accounting for interactions between these units provides an incomplete explanation. These interactions can be captured through circuit discovery methods. To address this, we propose a novel approach that enables joint pruning of connection edges and weight parameters, facilitating efficient neuron-level granularity. As a result, our method surpasses previous approaches, achieving state-of-the-art results in LM interpretation. Finally, our findings reveal two key insights for applications: global representational techniques that involve modifications to the entire model should be prioritised, while conventional, pre-theoretical speculation from cognitive science and linguistics has to date not been a fruitful enterprise. To demonstrate this, we present a case study on the task of temporal relation extraction, where, by leveraging these insights, we propose two effective improvements that achieve state-of-the-art performance.", "source_url": "https:\/\/openalex.org\/W7133020277", "pdf_url": "https:\/\/utoronto.scholaris.ca\/bitstreams\/bf75fbd3-79b0-4850-ab50-e543a20674e3\/download", "doi": "", "handle": "", "repository_name": "TSpace (University of Toronto)", "source_record_id": "W7133020277", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 396, "title": "Arbitrary-order Sampling and Hand Motion Modeling with Transformers", "title_ar": "", "title_en": "Arbitrary-order Sampling and Hand Motion Modeling with Transformers", "author": "Maxwell Clarke", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "New Zealand", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; responsible-ai; self-semi-supervised; transformers; fairness", "keywords": "", "abstract": "Transformer models have been quickly taking over many tasks in the field of deep learning, due to their versatility while maintaining high performance. This thesis provides an introduction to transformer models, presents experiments with new ways of sampling data with them, and then applies them to the domain of hand motion modeling. Firstly, a comprehensive introduction to transformer models is given, including the attention operation, masking, architecture variants, and different pre-training tasks. Secondly, an experiment is presented using a probabilistic transformer model on the MNIST dataset, which was trained so that it is capable of arbitrary-order sampling. The experiment compares different sampling orders, including some dynamic sampling order heuristics based on the entropy. The experiments find that such sampling orders introduce a statistical bias into the samples. Lastly, the problem domain of hand motion modeling is introduced, and transformer models are trained to generate hand-motion sequences, via self-supervised learning on a motion-capture dataset. Both deterministic and probabilistic models are trained. The deterministic models can generate realistic-looking hand motions, but cannot be directed to generate specific motions. The probabilistic model performs poorly.", "source_url": "https:\/\/openalex.org\/W4389570359", "pdf_url": "https:\/\/openaccess.wgtn.ac.nz\/articles\/thesis\/Arbitrary-order_Sampling_and_Hand_Motion_Modeling_with_Transformers\/24785205\/1\/files\/43587534.pdf", "doi": "10.26686\/wgtn.24785205", "handle": "", "repository_name": "Open Access Victoria University of Wellington | Te Herenga Waka (Figshare)", "source_record_id": "W4389570359", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 399, "title": "Zero\/Few shot learning applied to text classification", "title_ar": "", "title_en": "", "author": "Quang Anh Nguyen", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2026, "language": "fr", "research_domain": "nlp", "research_subdomain": "few-zero-shot", "categories": "machine-learning; computer-vision; nlp; ai-applications; few-zero-shot; ai-industry", "keywords": "", "abstract": "Zero\/Few-shot learning appliqué à la classification de textes Les institutions financières s'appuient de plus en plus sur des données textuelles non structurées, pour surveiller le risque de crédit. Leur volume et l'hétérogénéité rendent l'analyse manuelle complexe, d'où la classification automatique de texte devient essentielle. Cette thèse, réalisée dans le cadre du projet GoldenEye du Groupe BPCE, traite des défis liés à la limitation des données annotées et aux grands espaces d'étiquettes évolutifs. L'annotation manuelle est coûteuse et impratique à grande échelle, ce qui motive l'adoption des approches zero-shot et few-shot, réduisant la dépendance aux données annotées tout en maintenant des performances robustes. L'objectif principal est de concevoir des méthodes de classification de texte efficaces en termes de données, adaptées aux contraintes industrielles et aux contextes de supervision limitée. Plutôt que de se concentrer sur un seul modèle, la thèse explore plusieurs axes complémentaires en zero-shot et few-shot.La première contribution porte sur la classification few-shot multi-classe avec des modèles de langage masqués basés sur des prompts. Les verbalisateurs, correspondance entre classes et ensembles de mots-étiquettes, sont essentiels pour relier les étiquettes aux prédictions de tokens du modèle. Cette thèse propose un cadre de benchmarking pour les verbalizers manuels, automatiques et soft, montrant que le choix du verbalisateur influence fortement la performance en few-shot. Basé sur cette étude MaVEN (Manual Verbalizer Enrichment with Nearest Neighbors' Embedding) enrichit les verbalisateurs manuels en utilisant les voisins dans l'espace d'embedding, améliorant la performance.La deuxième contribution concerne la classification few-shot multi-étiquette, où la dépendance entre étiquettes posent des défis aux méthodes traditionnelles traitant chaque étiquette indépendamment. La thèse propose une formulation séquence-à-ensemble avec des transformeurs texte-à-texte (T5), considérant la prédiction multi-étiquette comme une génération séquentielle des prédictions pour capturer cette dépendance. Un pipeline d'entraînement en trois étapes (pré-entraînement, adaptation au domaine et spécialisation des chaînes) permet une forte généralisation inter-domaines. Les expériences montrent des performances supérieures aux baselines, et l'analyse d'ablation interprète les caractéristiques de fonctionnement de la méthode proposée.La troisième contribution traite la classification zero-shot multi-étiquette extrême, avec des espaces d'étiquettes très grands et aucune donnée annotée. La thèse propose un cadre d'auto-entraînement contrastif guidé par curriculum, qui raffine progressivement les représentations en utilisant des tâches de cloze inversé informé d'étiquettes, exploitant la sémantique tout en contrôlant le bruit des pseudo-étiquettes. Les expériences sur des benchmarks à grande échelle montrent des performances compétitives par rapport à de baselines zero-shot. L'analyse souligne que la conception attentive de l'entraînement permet un apprentissage zero-shot efficace même à grande échelle.En faisant progresser la classification zero-shot et few-shot, tant sur le plan méthodologique que pratique, ce travail montre que la formulation des tâches (via verbalizers, génération séquentielle et conception de curriculum) est un levier clé dans l'apprentissage sous supervision limitée. Ancrée dans un contexte industriel, la thèse fait le lien entre la recherche académique et le déploiement en conditions réelles.", "source_url": "https:\/\/openalex.org\/W7165153560", "pdf_url": "", "doi": "10.70675\/b574bae2zdc8dz4ce0za1cfzc369a572bfee", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7165153560", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 401, "title": "Analisis Performa Zero-shot dan Few-shot Learning Model Generative AI Dibandingkan dengan Fine-Tuning Model Berbasis BERT dalam Klasifikasi Emosi Ulasan Produk E-commerce Berbahasa Indonesia=Performance Analysis of Zero-Shot and Few-Shot Learning in Generative AI Models Compared with Fine-Tuning BERT-Based Models for Emotion Classification of Indonesian E-Commerce Product Reviews", "title_ar": "", "title_en": "", "author": "Rasyad Bimasatya", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Indonesia", "governorate": "", "publication_year": 2026, "language": "id", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; generative-ai; transfer-learning; few-zero-shot; transformers", "keywords": "", "abstract": "Latar belakang. Pesatnya pertumbuhan e-commerce di Indonesia meningkatkan volume ulasan produk emosional yang sulit dianalisis secara manual. Meskipun model BERT hasil fine-tuning terbukti efektif, pendekatannya memerlukan banyak data berlabel dan sumber daya komputasi. Sebaliknya, Generative AI menawarkan fleksibilitas melalui zero-shot dan few-shot learning, namun performanya pada ulasan berbahasa Indonesia masih perlu dievaluasi. Tujuan. Penelitian ini bertujuan untuk menganalisis sejauh mana performa teknik zero-shot dan few-shot learning pada model Generative AI dalam klasifikasi emosi ulasan produk e-commerce berbahasa Indonesia, serta membandingkannya dengan model berbasis BERT hasil fine-tuning, baik dari sisi efektivitas maupun efisiensi. Metode. Penelitian menggunakan pendekatan kuantitatif dengan mengevaluasi performa model menggunakan metrik precision, recall, accuracy, F1 score, latency, dan throughput. Selain itu, Pearson’s correlation coefficient digunakan untuk menganalisis hubungan antara ukuran few-shot learning dan perubahan performa model Generative AI. Model terbaik selanjutnya diimplementasikan ke dalam sistem berbasis website. Hasil. IndoBERT hasil fine-tuning mencapai performa tertinggi dengan accuracy dan F1 score macro average sebesar 84%, serta efisiensi yang lebih baik dibandingkan model Generative AI. Penambahan contoh few-shot meningkatkan performa beberapa model, namun menurunkan efisiensi. Kesimpulan. Penelitian ini menunjukkan bahwa performa teknik zero-shot dan few-shot learning pada model Generative AI berada di bawah model berbasis BERT hasil fine-tuning dalam tugas klasifikasi emosi ulasan e-commerce berbahasa Indonesia. Model IndoBERT dengan pendekatan fine-tuning telah diimplementasikan dalam aplikasi berbasis Streamlit sebagai solusi praktis analisis emosi ulasan.", "source_url": "https:\/\/openalex.org\/W7217542476", "pdf_url": "https:\/\/repository.unhas.ac.id\/id\/eprint\/57569\/2\/H071221024-1-2.pdf", "doi": "", "handle": "", "repository_name": "Hasanuddin University Repository", "source_record_id": "W7217542476", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 404, "title": "Attribute-based analysis of zero-shot learners", "title_ar": "", "title_en": "Attribute-based analysis of zero-shot learners", "author": "Vogel, Felix", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Goethe University Frankfurt", "university_source": "repository_host", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; few-zero-shot; multimodal; llm", "keywords": "", "abstract": "Vision-language models trained on large, randomly collected data had a significant impact in many areas since they appeared. But although they show great performance in various fields, such as image-text-retrieval, their inner workings are still not fully understood. An interesting application of those models is the zero-shot scenario that is analyzed in this work. A new benchmark consisting of five experiment setups is proposed to evaluate how zero-shot learners handle class labels and attributes. The experiments are divided into two sections, the classification performance with class label containing the first three experiment setups and the attribute based classification performance containing the last two experimental setups. The suggested experiments rely on attribute-based datasets with attribute annotations per image. Since only a few datasets with image attribute annotations exist, this work proposes a method to generate image attribute annotations. For a dataset that contains attribute annotations for classes, image attribute annotations are created with the help of a vision-language (VL) model. To validate if current zero-shot learners are really capable of zero-shot, an occurrence analysis is suggested. The recently released LAION-400M data corpus and the publicly available OpenCLIP model trained on LAION-400M are leveraged and ImageNet, AWA2, and CUB are used as evaluation datasets. For each class of an evaluation dataset, the training samples corresponding to the evaluation class are counted and the correlation between the model performance and class occurrences in the training data is provided. Using the assigned image attributes for the AWA2 and the CUB dataset, the proposed benchmark is applied to three publicly available VL models (OpenCLIP, CLIP, FLAVA). With a different prompt generation technique, the five experiment setups are repeated for AWA2 which leads to similar results and indicates their robustness. The provided analysis shows that: • Most of the classes in popular zero-shot benchmarks are observed (a lot) during pretraining. • Zero-shot performance mainly comes out of models’ capability of recognizing class labels whenever they are present in the text, and a significantly lower performing capability of attribute-based zero-shot learning is only observed when class labels are not used. • The number of attributes used can have a significant effect on performance, and can easily cause a significant performance decrease.", "source_url": "https:\/\/openalex.org\/W7112135146", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Publication Server of Goethe University Frankfurt am Main (Goethe University Frankfurt)", "source_record_id": "W7112135146", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 405, "title": "Zero- And Few-Shot Dark Kinase–Phosphosite Prediction Via Task-Aware Protein Embeddings", "title_ar": "", "title_en": "Zero- And Few-Shot Dark Kinase–Phosphosite Prediction Via Task-Aware Protein Embeddings", "author": "Zeynep Işık", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Turkey", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; nlp; medical-ai; transfer-learning; few-zero-shot; predictive-analytics; transformers; llm", "keywords": "", "abstract": "Accurately mapping kinases to their substrate phosphosites is fundamental fordecoding cellular signaling and understanding disease mechanisms. While highthroughputtechniques can identify the phosphosites, finding the kinase that catalyzesthe phosphorylation is challenging. Thus, over 95% of experimentally detectedhuman phosphosites lack kinase annotations. It is possible to formulatethe kinase-phosphosite association problem as a supervised multi-class classificationtask; however, a large portion of the human kinases are under-studied (darkkinases) and have few or no phosphosites associated with them, thus dark kinasesfall outside the reach of conventional supervised learning methods. In this thesis, weformulate kinase–phosphosite association as zero-shot and few-shot learning tasks:in the zero-shot setting, the model must predict associations for kinases never seenduring training; in the few-shot setting, it may leverage only a handful of labeledexamples.We employ transformer-based protein language models (pLMs) to embed both kinasedomains and phosphosite peptides, and we systematically explore domainadaptationstrategies—ranging from full fine-tuning and partial layer re-initializationto task-specific pre-training—under severe data constraints. Surprisingly, a denovo–trained ESM-1b model outperforms its fully fine-tuned pretrained counterpart,suggesting that general-purpose pLM embeddings may lack task-specific biochemicalcontext. Our best results are obtained by combining kinase- and phosphosite-aware pLMs with partial re-initialization of upper transformer layers. On the DARKINbenchmark, this approach delivers state-of-the-art performance in both zero-shotand few-shot kinase prediction, offering a promising direction for illuminating thedark phosphoproteome.", "source_url": "https:\/\/openalex.org\/W7124521015", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Sabanci University", "source_record_id": "W7124521015", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 406, "title": "LLMs for Structured Educational Data: A Zero and Few-Shot Approach to Student Performance Prediction", "title_ar": "", "title_en": "LLMs for Structured Educational Data: A Zero and Few-Shot Approach to Student Performance Prediction", "author": "Karandeep Singh Mann", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "National College of Ireland", "university_source": "repository_host", "college": "", "department": "", "country": "Ireland", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; nlp; responsible-ai; ai-applications; transfer-learning; few-zero-shot; predictive-analytics; llm; xai; ai-education", "keywords": "", "abstract": "Early and accurate prediction of student performance is crucial for enabling timely academic support to the at-risk students. The past work have used machine learning and deep learning approaches on educational datasets such as OULAD and achieved strong results for prediction of student outcomes. But they relied on feature engineering, required retraining and did not offered any interpretability. This study tries to check whether large language models (LLMs) can be used to overcome these limitations by using zero and few-shot prompting strategies on structured tabular datasets. Evaluation has been done on three open-source LLMs across binary, multiclass and per-class settings. The models reached up to accuracies of 89.6% and 75.9% on binary and multi-class predictions respectively, all without the need for fine-tuning. They even produced short natural language explanations to describe why a particular prediction was made. The findings show that LLMs can not only become an alternative to the traditional prediction models but can also guide educators to make informed decisions through their explanation based insights.", "source_url": "https:\/\/openalex.org\/W7163607681", "pdf_url": "https:\/\/norma.ncirl.ie\/9331\/2\/karandeepsinghmannconfigurationmanual.pdf", "doi": "", "handle": "", "repository_name": "TRAP@NCI (National College of Ireland)", "source_record_id": "W7163607681", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 411, "title": "Deep Zero- and Few-shot Learning in Computer Vision", "title_ar": "", "title_en": "Deep Zero- and Few-shot Learning in Computer Vision", "author": "Hongguang Zhang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Australian National University", "university_source": "repository_host", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; deep-learning; computer-vision; generative-ai; transfer-learning; few-zero-shot; cnn; object-detection; video-analysis; gan", "keywords": "", "abstract": "Most CNN models rely on the large-scale annotated training data, and the performance turns to be low when the size of training data is limited, therefore numbers of large-scale datasets are proposed for vision tasks. Though training on large-scale dataset can significantly improve the performance, creating such datasets for novel scenarios is costly. Moreover, relying on the number of samples is not consistent with human's learning abilities, as humans have the ability to understand novel concepts from limited examples. Transfer learning is a machine learning topic that addresses the above limitations in current deep models, and it studies how to make machines exploit the acquired knowledge to solve new problems, thus learning by generalizing from existing to new concepts, a trait akin to humans. Empirically, transfer learning for computer vision problems includes the following subtopics: i) zero-shot learning, ii) few-shot learning, and iii) domain adaptation. In this thesis, we study zero-shot and few-shot learning to demonstrate how to learn better from limited training samples. Zero-shot learning requires learning a mapping that associates feature vectors extracted from images with semantic annotations that describe objects and\/or scenes of interest. Our first work addresses the weakness of linear mapping in previous zero-shot learning models by learning via non-linear kernelized projections between features and attribute spaces to promote accuracy. We propose an easy learning objective inspired by the Linear Discriminant Analysis method with an incoherence mechanism. The dominant challenge in the generalized zero-shot setting is the unbalanced data distribution during training and testing, which makes it difficult for the classifier to distinguish if a testing sample is from a seen or unseen class. Inspired by GAN-based models, we propose the model selection mechanism, in which we leverage two sources of datapoints to train the model selection classifier to recognize which test datapoints come from seen and which from unseen classes. This way, generalized zero-shot learning can be decoupled into two disjoint classification tasks. Few-shot learning is a variant of knowledge transfer which is formulated as a meta-learning process. Firstly we investigate so-called second-order pooling and Power Normalizations in fine-grained classification tasks to study a proper way of using co-occurrence statistics for such tasks. Following above experiments and analysis, we propose the novel second-order similarity network, which measures image relations via co-occurrence relation descriptors, and significantly improves the performance of few-shot learning. Hallucinating auxiliary samples for training is also a popular and effective way to improve the accuracy of few-shot learning. To this end, we propose a saliency-guided hallucination strategy, which employs a saliency detector to segment foregrounds and backgrounds from sampled images forming an episode, then we mix every foreground with all available backgrounds in convolutional feature space to synthesize descriptors representing objects in various scenes. Furthermore, we note that previous few-shot learning methods fail to address the scale- and location-mismatch between support-query objects, thus leading to the performance loss. Inspired by the spatial-pyramid matching, we propose a novel spatial- and scale-matching network to explicitly match the support-query pairs over different scales and locations. Last but not least, we consider a more challenging scenario, namely few-shot action recognition, to study knowledge transfer in the video domain. Apart from the spatial information, robust aggregation over the temporal mode is fundamental in this task. To address this issue, we propose a novel action relation network with permutation-invariant attentions and auxiliary self-supervision tasks.", "source_url": "https:\/\/openalex.org\/W3112219073", "pdf_url": "", "doi": "10.25911\/5f5a020a5c26f", "handle": "", "repository_name": "ANU Open Research (Australian National University)", "source_record_id": "W3112219073", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 412, "title": "Zero-shot Learning: Recognition, Tagging, and Detection of Novel Concepts", "title_ar": "", "title_en": "Zero-shot Learning: Recognition, Tagging, and Detection of Novel Concepts", "author": "Shafin Rahman", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Australian National University", "university_source": "repository_host", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; responsible-ai; few-zero-shot; predictive-analytics; cnn; object-detection; medical-imaging; fairness", "keywords": "", "abstract": "Recent advancements in deep neural networks have performed favourably well on the supervised object recognition task. Towards an ultimate automated visual recognition system, we identify three key shortcomings of the existing supervised learning approaches. First, the dependency on a significantly large volume of manually annotated examples (e.g., ImageNet dataset with ~10 million images) limits the scalability of deep networks. Second, once the model learning stage is complete, it is difficult to add new classes continually when new data becomes available. Lastly, such models lack the notion of human-like understanding, i.e., an object can be recognized by humans without having any visual examples and just by a semantic description of its distinctive characteristics. In this thesis, we investigate the zero-shot learning (ZSL) framework to address these limitations. ZSL aims to perform reasoning about previously unseen objects without observing even a single instance of them. Such a learning paradigm requires no visual examples of novel objects, no re-training to add new classes and does not rely solely on visual information. Considering the relationship among the semantic description of previously seen examples, ZSL incorporates human wisdom to the visual understanding developed by a machine. In this work, we specifically address three critical bottlenecks in ZSL research which give rise to three ZSL problem settings: (a) unified zero-shot recognition (ZSR), (b) zero-shot tagging (ZST), and (c) zero-shot detection (ZSD) of novel concepts. Established ZSR methods are not flexible enough to adapt to one\/few-shot learning (O\/FSL) scenario where one\/few labeled examples of unseen classes become available during the supervised learning stage. To provide a comprehensive and flexible solution, we present a novel 'unified' approach for ZSL and O\/FSL based on class adapting principal direction (CAPD) that computes class-specific discriminative information by relating the semantic description of categories. The primary objective is to learn a metric in the semantic embedding space that minimizes intra-class distances and maximizes inter-class distances. In the real-life scenario, instead of a single object per image, a scene may contain multiple seen and unseen concepts together. To adopt this, we present the Deep0Tag approach for zero-shot tagging (ZST) to assign multiple labels to an input. This method considers both global and local details to discover seen or unseen concepts from a given scene. We solve this problem by formulating a multiple instance learning (MIL) framework. Unlike traditional MIL solutions, our method runs end-to-end without using offline object proposal generation methods. While most of the ZSL methods provide answers to unseen categories in simple tasks, e.g., single or multi-label classification and retrieval, we also focus on predicting both multi-class category-label and precise location of each instance in a given image. To this end, we introduce a new challenge for ZSL called zero-shot detection (ZSD) that simultaneously recognizes and localizes multiple novel concepts. Similar to traditional object detection, we present zero-shot version of double (Faster R-CNN) and single (RetinaNet) stage end-to-end object detectors. For both of the cases, we design associated loss functions that consider visual-semantic relationships to train the network. In addition to inductive learning approaches, we also propose the first transductive learning method for ZSD to reduce the domain-shift and model-bias against unseen classes convincingly. Our transductive approach follows a self-learning mechanism that uses a novel hybrid pseudo-labeling technique. Finally, we recommend training and testing protocols to evaluate ZSD based on large-scale ILSVRC-2017 and MSCOCO-2014 datasets.", "source_url": "https:\/\/openalex.org\/W3036608216", "pdf_url": "", "doi": "10.25911\/5ee74e8179e71", "handle": "", "repository_name": "ANU Open Research (Australian National University)", "source_record_id": "W3036608216", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 413, "title": "Zero-shot Image Classification", "title_ar": "", "title_en": "Zero-shot Image Classification", "author": "Yang Long", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "computer-vision", "research_subdomain": "image-classification", "categories": "machine-learning; computer-vision; generative-ai; few-zero-shot; image-classification", "keywords": "", "abstract": "Image classification is one of the essential tasks for the intelligent visual system. Conventional image classification techniques rely on a large number of labelled images for supervised learning, which requires expensive human annotations. Towards real intelligent systems, a more favourable way is to teach the machine how to make classification using prior knowledge like humans. For example, a palaeontologist could recognise an extinct species purely based on the textual descriptions. To this end, Zero-Shot Image Classification (ZIC) is proposed, which aims to make machines that can learn to classify unseen images like humans. The problem can be viewed from two different levels. Low-level technical issues are concerned by the general Zero-shot Learning (ZSL) problem which considers how to train a classifier on the unseen visual domain using prior knowledge. High-level issues incorporate how to design and organise visual knowledge representation to construct a systematic ontology that could be an ultimate knowledge base for machines to learn. This thesis aims to provide a thorough study of the ZIC problem, regarding models, challenges, possible applications, etc. Besides, each main chapter demonstrates an innovative contribution that is creatively made during my study. The first is to solve the problem of Visual-Semantic Ambiguity. Namely, the same semantic concepts (e.g. attributes) can refer to a huge variety of visual features, and vice versa. Conventional ZSL methods usually adopt a one-way embedding that maps such high-variance visual features into the semantic space, which may lead to degraded performance. As a solution, a dual-graph regularised embedding algorithm named Visual-Semantic Ambiguity Removal (VSAR) is proposed, which can capture the intrinsic local structure of both visual and semantic spaces. In the intermediate embedding space, the structural difference is reconciled to remove the ambiguity. The second contribution aims to circumvent costly visual data collection for conventional supervised classification using ZSL techniques. The key idea is to synthesise visual features from the semantic information, just like humans can imagine features of an unseen class from the semantic description of prior knowledge. Hereafter, new objects from unseen classes can be classified in a conventional supervised framework using the inferred visual features. To overcome the correlation problem, we propose an intermediate Orthogonal Semantic-Visual Embedding (OSVE) space to remove the correlated redundancy. The proposed method achieves promising performance on fine-grained datasets. In the third contribution, the graph constraint of VSAR is incorporated to synthesise improved visual features. The orthogonal embedding is reconsidered as an Information Diffusion problem. Through an orthogonal rotation, the synthesised visual features become more discriminative. On four benchmarks, the proposed method demonstrates the advantages of synthesised visual features, which significantly outperforms state-of-the-art results. Since most of ZSL approaches highly rely on expensive attributes, the fourth contribution of this thesis explores a more feasible but more effective Semantic Simile model to describe unseen classes. From a group of similes, e.g. an unknown animal has the same parts of a wolf, and the colour looks like a bobcat, implicit attributes are discovered by a graph-cut algorithm. Comprehensive experimental results suggest the simile-based implicit attributes can significantly boost the performance. To maximumly reduce the cost of building ontologies for ZIC, the final chapter introduces a novel scheme, using which ZIC can be achieved by only a few similes of each unseen class. No annotations of seen classes are needed. Such an approach finally sets ZIC attribute-free, which significantly improve the feasibility of ZIC. Unseen classes can be recognised using a conventional setting without expensive attribute ontology. It can be concluded that the methods introduced in this thesis provide fundamental components of a zero-shot image classification system. The thesis also points out four core directions for future ZIC research.", "source_url": "https:\/\/openalex.org\/W2770447235", "pdf_url": "", "doi": "", "handle": "", "repository_name": "White Rose eTheses Online (University of Leeds, The University of Sheffield, University of York)", "source_record_id": "W2770447235", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 418, "title": "Low-shot Learning based on Pre-trained Model", "title_ar": "", "title_en": "Low-shot Learning based on Pre-trained Model", "author": "Ziqin Zhou", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Adelaide", "university_source": "repository_host", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; responsible-ai; transfer-learning; few-zero-shot; self-semi-supervised; cnn; transformers; multimodal; image-classification; image-segmentation; fairness", "keywords": "", "abstract": "In the field of computer vision, deep learning has demonstrated remarkable accomplishments. However, the dependence on extensive annotated datasets poses significant challenges, particularly when such data is scarce, costly, or ethically sensitive. This thesis addresses the crucial issue of low-shot learning by aiming to equip learning models with the capability to generalize effectively from limited examples. It explores various low-supervision methodologies, including self-supervised, unsupervised, semi-supervised, and active learning, with the goal of reducing the reliance on labeled data. One potential solution to this challenge is leveraging pre-trained models trained on large-scale datasets and transferring them to downstream tasks and unknown classes. Nevertheless, directly applying pre-trained models in a frozen state often fails to meet the requirements of specific tasks. Fine-tuning these models on base classes with limited training samples and then transferring them to novel domains presents obstacles, as it may disrupt the general knowledge encoded in the model, leading to overfitting on the base classes and suboptimal performance on unseen categories. The central focus of this research is to fully exploit the knowledge encoded in large pre-trained models and enhance their capacity to transfer to novel classes. This thesis introduces a systematic approach to balance generalization and task-specific adaptation, ensuring that fine-tuning preserves pre-trained knowledge while effectively adapting to new data. These models encapsulate extensive knowledge from diverse data sources, enabling swift adaptation to new tasks and classes with minimal additional training. The downstream tasks investigated in this doctoral work range from the low-shot problem for few-shot image classification to more intricate endeavors such as zero-shot and few-shot pixel-level semantic segmentation. The exploration encompasses various pre-trained model architectures, including convolutional and transformer-based models, spanning both pure vision and cross-modal vision-language paradigms. By harnessing the representational capabilities of state-of-the-art pre-trained models, this research aims to push the boundaries of what is achievable with limited data. The novel methodologies proposed strive to maximize performance, maintain generalization across domains, and improve inference efficiency, paving the way for broader adoption in low-data scenarios.", "source_url": "https:\/\/openalex.org\/W7216837125", "pdf_url": "https:\/\/digital.library.adelaide.edu.au\/server\/api\/core\/bitstreams\/c8d7f0de-908a-4f17-b259-cc6fb0be75e4\/content", "doi": "", "handle": "", "repository_name": "Adelaide Research & Scholarship (AR&S) (University of Adelaide)", "source_record_id": "W7216837125", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 419, "title": "On Zero-Shot Reinforcement Learning", "title_ar": "", "title_en": "On Zero-Shot Reinforcement Learning", "author": "Scott Jeen", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "reinforcement-learning", "research_subdomain": "few-zero-shot", "categories": "machine-learning; reinforcement-learning; few-zero-shot", "keywords": "", "abstract": "Modern reinforcement learning (RL) systems capture deep truths about general, human problem-solving. In domains where new data can be simulated cheaply, these systems uncover sequential decision-making policies that far exceed the ability of any human. Society faces many problems whose solutions require this skill, but they are often in domains where new data cannot be cheaply simulated. In such scenarios, we can learn simulators from existing data, but these will only ever be approximately correct, and can be pathologically incorrect when queried outside of their training distribution. As a result, a misalignment between the environments in which we train our agents and the real-world in which we wish to deploy our agents is inevitable. Dealing with this misalignment is the primary concern of zero-shot reinforcement learning, a problem setting where the agent must generalise to a new task or domain with zero practice shots. Whilst impressive progress has been made on methods that perform zero-shot RL in idealised settings, new work is needed if these results are to be replicated in real-world settings. In this thesis, we argue that doing so requires us to navigate (at least) three constraints. First, the data quality constraint: real-world datasets are small and homogeneous. Second, the observability constraint: states, dynamics and rewards in the real-world are often only partially observed. And third, the data availability constraint: a priori access to data cannot always be assumed. This work proposes a suite of methods that perform zero-shot RL subject to these constraints. In a series of empirical studies we expose the failings of existing methods, and justify our techniques for remedying them. We believe these designs take us a step closer to RL methods that can be deployed to solve real-world problems.", "source_url": "https:\/\/openalex.org\/W4417060277", "pdf_url": "https:\/\/arxiv.org\/pdf\/2508.16496", "doi": "10.17863\/cam.121637", "handle": "", "repository_name": "arXiv (Cornell University)", "source_record_id": "W4417060277", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 430, "title": "Natural Language Processing in Finance and Empirical Macroeconomics", "title_ar": "", "title_en": "Natural Language Processing in Finance and Empirical Macroeconomics", "author": "Trebbi, Giovanni", "supervisor": "", "degree_type": "phd", "degree_evidence": "نص السجل: «Ph.D»", "university": "Tilburg University", "university_source": "repository_host", "college": "", "department": "", "country": "Netherlands", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "predictive-analytics", "categories": "machine-learning; nlp; predictive-analytics", "keywords": "", "abstract": "The three essays collected in this Ph.D. thesis apply Natural Language Processing (NLP) tools to asset pricing and household finance. The first essay identifies the triggers of flight-to-safety episodes from financial news and highlights the importance of political, geopolitical, and macroeconomic risks. The second essay shows how the inflation expectation gap between households and experts widens with narrative disagreement between general and specialized newspapers about the demand or supply origins of inflation. The third essay demonstrates how the temporal references of newspaper articles can add predictive power toward future excess stock returns above and beyond their tone and topics.", "source_url": "https:\/\/openalex.org\/W7103976642", "pdf_url": "", "doi": "10.26116\/tisem.56616372", "handle": "", "repository_name": "Research portal (Tilburg University)", "source_record_id": "W7103976642", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 434, "title": "Word Embeddings for Natural Language Processing", "title_ar": "", "title_en": "Word Embeddings for Natural Language Processing", "author": "Rémi Lebret", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Ecole Polytechnique Fédérale de Lausanne", "university_source": "repository_host", "college": "", "department": "", "country": "Switzerland", "governorate": "", "publication_year": 2016, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; data-mining; predictive-analytics; cnn; llm; sentiment-analysis", "keywords": "", "abstract": "Word embedding is a feature learning technique which aims at mapping words from a vocabulary into vectors of real numbers in a low-dimensional space. By leveraging large corpora of unlabeled text, such continuous space representations can be computed for capturing both syntactic and semantic information about words. Word embeddings, when used as the underlying input representation, have been shown to be a great asset for a large variety of natural language processing (NLP) tasks. Recent techniques to obtain such word embeddings are mostly based on neural network language models (NNLM). In such systems, the word vectors are randomly initialized and then trained to predict optimally the contexts in which the corresponding words tend to appear. Because words occurring in similar contexts have, in general, similar meanings, their resulting word embeddings are semantically close after training. However, such architectures might be challenging and time-consuming to train. In this thesis, we are focusing on building simple models which are fast and efficient on large-scale datasets. As a result, we propose a model based on counts for computing word embeddings. A word co-occurrence probability matrix can easily be obtained by directly counting the context words surrounding the vocabulary words in a large corpus of texts. The computation can then be drastically simplified by performing a Hellinger PCA of this matrix. Besides being simple, fast and intuitive, this method has two other advantages over NNLM. It first provides a framework to infer unseen words or phrases. Secondly, all embedding dimensions can be obtained after a single Hellinger PCA, while a new training is required for each new size with NNLM. We evaluate our word embeddings on classical word tagging tasks and show that we reach similar performance than with neural network based word embeddings. While many techniques exist for computing word embeddings, vector space models for phrases remain a challenge. Still based on the idea of proposing simple and practical tools for NLP, we introduce a novel model that jointly learns word embeddings and their summation. Sequences of words (i.e. phrases) with different sizes are thus embedded in the same semantic space by just averaging word embeddings. In contrast to previous methods which reported a posteriori some compositionality aspects by simple summation, we simultaneously train words to sum, while keeping the maximum information from the original vectors. These word and phrase embeddings are then used in two different NLP tasks: document classification and sentence generation. Using such word embeddings as inputs, we show that good performance is achieved in sentiment classification of short and long text documents with a convolutional neural network. Finding good compact representations of text documents is crucial in classification systems. Based on the summation of word embeddings, we introduce a method to represent documents in a low-dimensional semantic space. This simple operation, along with a clustering method, provides an efficient framework for adding semantic information to documents, which yields better results than classical approaches for classification. Simple models for sentence generation can also be designed by leveraging such phrase embeddings. We propose a phrase-based model for image captioning which achieves similar results than those obtained with more complex models. Not only word and phrase embeddings but also embeddings for non-textual elements can be helpful for sentence generation. We, therefore, explore to embed table elements for generating better sentences from structured data. We experiment this approach with a large-scale dataset of biographies, where biographical infoboxes were available. By parameterizing both words and fields as vectors (embeddings), we significantly outperform a classical model.", "source_url": "https:\/\/openalex.org\/W2540566554", "pdf_url": "", "doi": "10.5075\/epfl-thesis-7148", "handle": "", "repository_name": "Infoscience (Ecole Polytechnique Fédérale de Lausanne)", "source_record_id": "W2540566554", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 445, "title": "Natural language processing in cross-media analysis", "title_ar": "", "title_en": "Natural language processing in cross-media analysis", "author": "Yonas Demeke Woldemariam", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Umeå University", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "nlp", "research_subdomain": "sentiment-analysis", "categories": "machine-learning; deep-learning; computer-vision; nlp; predictive-analytics; multimodal; video-analysis; sentiment-analysis; speech-recognition", "keywords": "", "abstract": "A cross-media analysis framework is an integrated multi-modal platform where a media resource containing different types of data such as text, images, audio and video is analyzed with metadata extractors, working jointly to contextualize the media resource. It generally provides cross-media analysis and automatic annotation, metadata publication and storage, searches and recommendation services. For on-line content providers, such services allow them to semantically enhance a media resource with the extracted metadata representing the hidden meanings and make it more efficiently searchable. Within the architecture of such frameworks, Natural Language Processing (NLP) infrastructures cover a substantial part. The NLP infrastructures include text analysis components such as a parser, named entity extraction and linking, sentiment analysis and automatic speech recognition. Since NLP tools and techniques are originally designed to operate in isolation, integrating them in cross-media frameworks and analyzing textual data extracted from multimedia sources is very challenging. Especially, the text extracted from audio-visual content lack linguistic features that potentially provide important clues for text analysis components. Thus, there is a need to develop various techniques to meet the requirements and design principles of the frameworks. In our thesis, we explore developing various methods and models satisfying text and speech analysis requirements posed by cross-media analysis frameworks. The developed methods allow the frameworks to extract linguistic knowledge of various types and predict various information such as sentiment and competence. We also attempt to enhance the multilingualism of the frameworks by designing an analysis pipeline that includes speech recognition, transliteration and named entity recognition for Amharic, that also enables the accessibility of Amharic contents on the web more efficiently. The method can potentially be extended to support other under-resourced languages.", "source_url": "https:\/\/openalex.org\/W2884821628", "pdf_url": "", "doi": "", "handle": "", "repository_name": "DiVA at Umeå University (Umeå University)", "source_record_id": "W2884821628", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 448, "title": "Automation of work processes using object detection in electrical installation images : Patch-based training and object detection", "title_ar": "", "title_en": "Automation of work processes using object detection in electrical installation images : Patch-based training and object detection", "author": "Alan Abdulrazzaq ; Björn Andersson", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "KTH Royal Institute of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "computer-vision; ai-applications; object-detection; ai-industry", "keywords": "", "abstract": "This thesis investigates how artificial intelligence and computer vision can be utilized to automate the identification of installation symbols in digital electrical drawings for the company mbiz AB. In the electrical installation industry, planning and selfmonitoring are currently performed largely manually, which is both time-consuming and labor-intensive. A key technical challenge is that the symbols are extremely small relative to the overall size and high resolution of the drawings, complicating automated analysis. The study evaluates a methodology based on the object detection model YOLOv11 in combination with SAHI (Slicing Aided Hyper Inference), a technique that divides the drawing into smaller segments to preserve detail. The results show that the model achieved a Mean Average Precision (mAP) of 0.97, confirming that patch-based inference is an effective method for handling technical documents. The study resulted in a functional prototype that integrates visual detection with automated reading of drawing numbers and coordinate translation to PDF format. The conclusion is that the technology has significant potential to increase efficiency and quality assurance in future installation projects by reducing manual tasks.", "source_url": "https:\/\/openalex.org\/W7132371682", "pdf_url": "", "doi": "", "handle": "", "repository_name": "KTH Publication Database DiVA (KTH Royal Institute of Technology)", "source_record_id": "W7132371682", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 450, "title": "Identification of errors in the 3D printing process using computer vision applications", "title_ar": "", "title_en": "Identification of errors in the 3D printing process using computer vision applications", "author": "Daniel Janovský", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Czech Republic", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; computer-vision; object-detection", "keywords": "", "abstract": "This thesis focuses on the integration and application of computer vision techniques in the field of 3D printing with the aim of automated detection of printing defects. The work analyzes the most common defects occurring during the 3D printing process and explores the possibilities of their identification using modern machine learning methods. A dataset for training the computer vision model was created based on data obtained from a database of images depicting various printing defects, which were subsequently categorized into specific classes. The dataset was manually annotated using the CVAT tool in COCO format, with individual defects labeled using polygonal annotations. In total, the dataset contains approximately 154 thousand annotated objects divided into five defect classes. Based on the prepared data, an object detection model utilizing the YOLOv12 architecture was trained, combining high inference speed with advanced image processing mechanisms. The model was trained using different hyperparameter settings, specifically with a fixed number of 100 epochs and a learning rate of 0.01, while the batch size was adjusted according to the input image resolution (256 for 320 px, 64 for 640 px, and 32 for 1080 px). Additionally, three input resolutions (320, 640, and 1080 px) were evaluated to optimize detection accuracy. The performance of the trained model was evaluated using standard metrics such as precision, recall, and mean Average Precision (mAP) at different IoU thresholds. The results showed that increasing the resolution led to an improvement in detection accuracy, with mAP50 increasing from approximately 0.63 (320 px) to 0.74 (640 px) and further to 0.77 (1080 px). A similar trend was observed for mAP50-95, which increased from 0.37 to 0.58, as well as for precision, which rose from 0.83 to 0.89, and recall, which improved from 0.56 to 0.73. The most significant improvement was observed when increasing the resolution from 320 px to 640 px.", "source_url": "https:\/\/openalex.org\/W7163860920", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Digital Repository (National Repository of Grey Literature)", "source_record_id": "W7163860920", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 451, "title": "Computer vision for non-rigid object assembly automation: With applications in automotive wire harness assembly", "title_ar": "", "title_en": "Computer vision for non-rigid object assembly automation: With applications in automotive wire harness assembly", "author": "Hao Wang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Chalmers University of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "computer-vision", "research_subdomain": "ai-industry", "categories": "deep-learning; computer-vision; robotics; ai-applications; multimodal; ai-industry", "keywords": "", "abstract": "This thesis examines the role of computer vision in facilitating robotic automation for non-rigid object assembly, with a particular focus on wire harness assembly tasks during the automotive final assembly stage. Despite extensive research in this field, industrial adoption has remained limited. Accordingly, the thesis analyzes the challenges associated with applying computer vision to wire harness assembly automation and explores strategies to enable practical deployment in industrial environments.Employing both qualitative and quantitative methods, the research progresses from problem identification to artifact design, demonstration, and evaluation in laboratory and industrially relevant environments. The studies identify challenges at the object, scene, data, and operational levels. To address these challenges, three primary artifacts were developed and evaluated. First, a learning-based perception pipeline enables markerless detection of wire harness components. This demonstrates the feasibility of deep learning-based component recognition and highlights limitations when components possess highly similar or occluded visual features. Second, a robot-assisted pipeline for automated multi-view data acquisition and multimodal annotation substantially accelerates the preparation of computer vision datasets. This pipeline also supports the training and evaluation of learning-based perception methods for industrial applications. Third, a vision-based human–robot collaboration framework for wire harness installation significantly reduces localized physical discomfort while maintaining task success. However, this approach increases mental demand and cycle time, with the majority of the additional time attributable to robot execution.In summary, this thesis provides deployable methods and practical guidance for data-centric development, interaction design, and takt-time-oriented workflow optimization in non-rigid object assembly automation. It also demonstrates that, given current technological constraints, computer vision is most effective as a human-centered enabler of robot-assisted assembly rather than as a direct pathway to fully autonomous robotic assembly of non-rigid objects.", "source_url": "https:\/\/openalex.org\/W7160860637", "pdf_url": "https:\/\/research.chalmers.se\/publication\/551870\/file\/551870_Fulltext.pdf", "doi": "10.63959\/chalmers.dt\/5880", "handle": "", "repository_name": "Chalmers Research (Chalmers University of Technology)", "source_record_id": "W7160860637", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 454, "title": "Object detection", "title_ar": "", "title_en": "Object detection", "author": "Pavel Vítek", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Czech Republic", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "computer-vision; object-detection", "keywords": "", "abstract": "This paper deals with object detection where a human figure is used as a model. The introductory chapter defines terms detection and object. The second chapter describes the basic theory of computer vision , such as image processing and its parts. A substantial part is also focused on the Canny edge detection, upper body detection and pose detection. The third chapter describes the MATLAB integrated development environment in which the image processing functions are implemented. The last chapter is a practical demonstration of object detection. Different upper body detection methods are compared there. In the last subsection, two methods of determining the deviations based on the template are compared. Finally, all methods are compared with respect to the obtained results.", "source_url": "https:\/\/openalex.org\/W7135601709", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Digital Repository (National Repository of Grey Literature)", "source_record_id": "W7135601709", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 457, "title": "Localization of UAVs Using Computer Vision in a GPS-Denied Environment", "title_ar": "", "title_en": "Localization of UAVs Using Computer Vision in a GPS-Denied Environment", "author": "Ram Charan Aluri", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; deep-learning; computer-vision; robotics; object-detection", "keywords": "", "abstract": "The main objective of this thesis is to propose a localization method for a UAV using various computer vision and machine learning techniques. It plays a major role in planning the strategy for the flight, and acts as a navigational contingency method, in event of a GPS failure. The implementation of the algorithms employs high processing capabilities of the graphics processing unit, making it more efficient. The method involves the working of various neural networks, working in synergy to perform the localization. This thesis is a part of a collaborative project between The University of North Texas, Denton, USA, and the University of Windsor, Ontario, Canada. The localization has been divided into three phases namely object detection, recognition, and location estimation. Object detection and position estimation were discussed in this thesis while giving a brief understanding of the recognition. Further, future strategies to aid the UAV to complete the mission, in case of an eventuality, like the introduction of an EDGE server and wireless charging methods, was also given a brief introduction.", "source_url": "https:\/\/openalex.org\/W4406532067", "pdf_url": "https:\/\/digital.library.unt.edu\/ark:\/67531\/metadc1944237\/m2\/1\/high_res_d\/ALURI-THESIS-2022.pdf", "doi": "10.12794\/metadc1944237", "handle": "", "repository_name": "University of North Texas Digital Library (University of North Texas)", "source_record_id": "W4406532067", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 458, "title": "Automated Curbside Waste Collection using Advanced Computer Vision Strategies", "title_ar": "", "title_en": "Automated Curbside Waste Collection using Advanced Computer Vision Strategies", "author": "Mohammad Ali Lordifar", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Guelph", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "deep-learning; computer-vision; robotics; ai-applications; cnn; object-detection; ai-industry", "keywords": "", "abstract": "Computer vision is an interesting and rapidly growing area of study that is often used in conjunction with artificial intelligence (AI) methods. The number of useful applications for computer vision techniques is vast, and includes object detection, analysis of human actions, autonomous vehicles, and target tracking.A local company has partnered with the Intelligent Control and Estimation Laboratory to work on further developing an automated waste collection system. Currently, the system utilizes an industrial-grade camera, mechatronics control box that integrates with waste collection trucks (e.g., robot arms), and customized AI software. The system relies heavily upon its ability to successfully and accurately detect a waste collection cart (e.g., garbage, recycling, or compost bin). The current methodology employs a significant amount of manual annotation or labelling of pictures to train the system to detect a cart; all of which is time consuming and costly. This research thesis implemented and further developed convolutional neural networks (CNN) and ‘you only look once’ (YOLO) object detection strategies. The application of CNN and YOLO to the automated waste collection system improved successful cart detection rates (about 98%), reduced false positives (less than 2%), and will improve operational safety and reliability when implemented in the real-time system.", "source_url": "https:\/\/openalex.org\/W3049386911", "pdf_url": "https:\/\/atrium.lib.uoguelph.ca\/bitstreams\/7abc16b0-85af-4178-a4a5-d0ec14beb1a9\/download", "doi": "", "handle": "", "repository_name": "The Atrium (University of Guelph)", "source_record_id": "W3049386911", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 459, "title": "Shoulder Keypoint-Detection from Object Detection", "title_ar": "", "title_en": "Shoulder Keypoint-Detection from Object Detection", "author": "Prince Kapoor", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Ottawa", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; deep-learning; computer-vision; ai-applications; predictive-analytics; cnn; rnn; object-detection; image-segmentation; ai-agriculture", "keywords": "", "abstract": "This thesis presents detailed observation of different Convolutional Neural Network (CNN) architecture which had assisted Computer Vision researchers to achieve state-of-the-art performance on classification, detection, segmentation and much more to name image analysis challenges. Due to the advent of deep learning, CNN had been used in almost all the computer vision applications and that is why there is utter need to understand the miniature details of these feature extractors and find out their pros and cons of each feature extractor meticulously. In order to perform our experimentation, we decided to explore an object detection task using a particular model architecture which maintains a sweet spot between computational cost and accuracy. The model architecture which we had used is LSTM-Decoder. The model had been experimented with different CNN feature extractor and found their pros and cons in variant scenarios. The results which we had obtained on different datasets elucidates that CNN plays a major role in obtaining higher accuracy and we had also achieved a comparable state-of-the-art accuracy on Pedestrian Detection Dataset. In extension to object detection, we also implemented two different model architectures which find shoulder keypoints. So, One of our idea can be explicated as follows: using the detected annotation from object detection, a small cropped image is generated which would be feed into a small cascade network which was trained for detection of shoulder keypoints. The second strategy is to use the same object detection model and fine tune their weights to predict shoulder keypoints. Currently, we had generated our results for shoulder keypoint detection. However, this idea could be extended to full-body pose Estimation by modifying the cascaded network for pose estimation purpose and this had become an important topic of discussion for the future work of this thesis.", "source_url": "https:\/\/openalex.org\/W2888395965", "pdf_url": "", "doi": "10.20381\/ruor-22271", "handle": "", "repository_name": "uO Research (University of Ottawa)", "source_record_id": "W2888395965", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 460, "title": "Advancing heat stress detection in dairy cows through machine learning and computer vision", "title_ar": "", "title_en": "Advancing heat stress detection in dairy cows through machine learning and computer vision", "author": "Shu, Hang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Liège", "university_source": "repository_host", "college": "", "department": "", "country": "Belgium", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; deep-learning; computer-vision; responsible-ai; ai-applications; predictive-analytics; cnn; object-detection; image-segmentation; face-recognition; video-analysis; xai; ai-agriculture", "keywords": "", "abstract": "Heat stress detection in dairy cows has long been connected with production loss. However, the reduction in milk yield lags behind the exposure to heat stress events for about two days. Other stress responses, such as physiological and behavioural changes, are well documented to be activated by dairy cows in the earlier stage of heat stress compared with production loss. Among all candidate indicators, body surface temperatures (BST), respiration rate (RR), and relevant behaviours have been concluded to be the most appropriate indicators due to their high feasibility of acquisition and early response. Vision-based methods are promising for accurate measurements while adhering to animal welfare principles. Meanwhile, predictive models show a non-invasive alternative to obtain these data and can provide useful insights with their interpretations. Thus, this thesis aimed to provide non-invasive solutions to the detection of heat stress in dairy cows by using artificial intelligence techniques. The detailed research content and relevant conclusions are as follows: An automated tool based on improved UNet was proposed to collect facial BST from five facial landmarks (i.e., eyes, muzzle, nostrils, ears, and horns) on cattle infrared images. The baseline UNet model was improved by replacing the traditional convolutional layers in the decoder with Ghost modules and adding efficient channel attention modules. The improved UNet outperformed other comparable models with the highest mean Intersection of Union of 80.76% and a slightly slower but still good inference speed of 32.7 frames per second (FPS). Agreement analysis reveals small to negligible differences between the temperatures obtained automatically in the area of eyes and ears and the ground truth. A vision-based method was proposed to measure RR for multiple dairy cows lying on free stalls. The proposed method involved various computer vision tasks (i.e., instance segmentation, object detection, object tracking, video stabilisation, and optical flow) to obtain respiration-related signals and finally utilised Fast Fourier Transform to extract RR. The results show that the measured RR had a Pearson correlation coefficient of 0.945, a root mean square error (RMSE) of 5.24 breaths per minute (bpm), and an intraclass correlation coefficient of 0.98 compared with visual observation. The average processing time and FPS on 55 test video clips (mean ± standard deviation duration of 16 ± 4 s) was 8.2 s and 64, respectively. A deep learning-based model was proposed to recognise cow behaviours (i.e., drinking, eating, lying, standing-in, and standing-out) that are known to be influenced by heat stress. The YOLOv5s model was selected due to its ability to compress the weight size while maintaining accuracy. It had a mean average precision of 0.985 and an inference speed of 73 FPS. Further validation demonstrates the excellent capacity of the proposed model in measuring herd-level behavioural indicators, with an intraclass correlation coefficient of 0.97 compared with manual observation. Critical thresholds were determined by using piecewise regression models with environmental indicators as the predictors and animal-based indicators as the outcomes. An ambient temperature (Ta) threshold was determined at 26.1 °C when the automated measured mean eye temperature reached 35.3 °C. A Ta threshold of 23.6 °C and a temperature-humidity index (THI) threshold of 72 were determined when the automated measured RR reached 61.1 and 60.4 bpm, respectively. In addition, the test dairy herd began to change their standing and lying behaviour at the earliest Ta of 23.8 ℃ or THI of 68.5. Four machine learning algorithms were used to predict RR, vaginal temperature (VT), and eye temperature (ET) from 13 predictor variables from three dimensions: production, cow-related, and environmental factors. The artificial neural networks yielded the lowest RMSE for predicting RR (13.24 bpm), VT (0.30 ℃), and ET (0.29 ℃). The results interpreted with partial dependence plots and Local Interpretable Model-agnostic Explanations show that P.M. measurements and winter calving contributed most to high RR and VT predictions, whereas lying posture, high Ta, and low wind speed contributed most to high ET predictions. Based on these results, an integrative application of all the proposed measurement, prediction, and assessment methods has been suggested, wherein RGB and infrared cameras are used to measure animal-based indicators, and critical thresholds, along with model interpretation, are used to assess the heat stress state of dairy cows. This strategy ensures timely and thorough cooling of cows in all areas of the dairy farm, thereby minimising the negative impact of heat stress to the greatest extent.", "source_url": "https:\/\/openalex.org\/W7111984551", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Open Repository and Bibliography (University of Liège)", "source_record_id": "W7111984551", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 463, "title": "Language Supervision for Computer Vision", "title_ar": "", "title_en": "Language Supervision for Computer Vision", "author": "Karan Desai", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Michigan", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; transfer-learning; few-zero-shot; multimodal; object-detection; image-segmentation; llm", "keywords": "", "abstract": "Representation learning lies at the core of modern Artificial Intelligence. In computer vision, labeled image datasets like ImageNet have been the standard choice for representation learning. Despite being empirically successful, this approach is expensive to scale due to labeling costs. Moreover, the representation quality is limited by the size and diversity of datasets and their associated label ontologies. My research explores using natural language supervision for computer vision. Using natural language allows us to go beyond fixed label ontologies and scale up to more general sources such as internet data. Toward this goal, my dissertation explores four problems -- (1) Learning representations: I propose one of the first methods for language-supervised visual learning that uses image captioning as the training objective, showing its efficacy compared to ImageNet-trained methods on downstream tasks like object detection and segmentation. (2) Scaling data: I explore social media as a rich source of high-quality image descriptions and curate a dataset of 12 million image-text pairs while ensuring responsible curation practices. (3) Understanding data: It is difficult to comprehend the diversity of visual concepts present in millions of image-text pairs. I posit that images and text naturally organize into a tree-like hierarchy and propose an approach for learning representations that capture this hierarchy using tools from hyperbolic geometry. (4) Transfer to downstream tasks: Large vision-language models show impressive zero-shot transfer capabilities on image-level tasks like classification and retrieval. However, their transferability to pixel-level tasks like object detection and segmentation has relied on expensive labeled mask annotations. I propose an object detector to efficiently transfer pre-trained vision models to segment and classify visual objects without any fine-tuning, unlike existing detectors that train using orders of magnitude more labeled masks to achieve high performance. In summary, my research affirms that using language supervision can drive the next leap of progress in computer vision and has immense utility in practical applications.", "source_url": "https:\/\/openalex.org\/W4398230580", "pdf_url": "", "doi": "10.7302\/22865", "handle": "", "repository_name": "Deep Blue (University of Michigan)", "source_record_id": "W4398230580", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 470, "title": "Deep Learning Object Detection Models in Robotic Process Automation", "title_ar": "", "title_en": "Deep Learning Object Detection Models in Robotic Process Automation", "author": "Waltteri Vuorimaa", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Aalto University", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; deep-learning; computer-vision; robotics; object-detection", "keywords": "", "abstract": "In recent years, businesses have become increasingly aware of their need to automate previously manual business processes. Adoption rates of robotic process automation (RPA) tools have increased a lot, and this has generated a great deal of value for companies. But, alas, especially some older legacy applications are still extremely problematic from the point of view of RPA platforms, and they cannot often be interfaced in a same fashion as some more modern applications. This is especially lamentable, as it is often the older applications, which tend to be used in the more manual business processes. This thesis aims to build a foundation for building and evaluating commercializable computer vision (CV) object detection models, that resolve the issues faced by RPA applications that arise from the lack of operating system (OS) level user interface (UI) visual element data. This kind of models could be used as a basis for building completely new RPA tools, or for augmenting existing solutions. The research questions discussed by this study are as follows: 1. Is using machine learning models for the detection of user interface elements viable? 2. What are the implications to existing RPA applications? 3. Can we build a model that is able to generalize from data from another domain? This thesis discusses the theoretical background of object detection and machine learning, and presents three potential families of machine learning model architectures that could be used for the aforementioned tasks. These models were then trained and compared to the CV capabilities of a commercial RPA platform, UiPath. The results show that the most simple object detection models cannot compete with the current commercial offering. There is, however, a clear path to further research, and it is clear that building a object detection system for UI elements is viable. In addition, this thesis presents the first publicly available dataset of (manually) annotated desktop UI screenshot images, and a tool for generating artificial user interface data.", "source_url": "https:\/\/openalex.org\/W3154899117", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Aaltodoc (Aalto University)", "source_record_id": "W3154899117", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 474, "title": "An efficient deep learning approach to detect plant diseases using image data", "title_ar": "", "title_en": "An efficient deep learning approach to detect plant diseases using image data", "author": "Ahmad Anis ; Babli Saha ; Maisa Tarannum Srizee", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "BRAC University", "university_source": "repository_host", "college": "", "department": "", "country": "Bangladesh", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; ai-applications; transfer-learning; predictive-analytics; cnn; transformers; ai-agriculture", "keywords": "", "abstract": "Plant diseases drastically impact on the yield and quality of crops and this is a great challenge to food security and sustainable agriculture in the world and more so in countries such as Bangladesh. There is need to therefore ensure the early and accurate diagnosis of plant diseases to enhance productivity and safeguard food supplies. Plant disease classification based on images has been highly successful with the swift development of deep learning, especially Convolutional Neural Networks (CNNs).In this thesis, a lightweight hybrid CNN-Transformer model is suggested as a classification of rice leaf disease. A number of pretrained CNN models, such as ResNet50, VGG16, InceptionV3, MobileNet V3, EfficientNetB0, and DenseNet121 are tested with transfer learning to determine the baseline performance. The dataset used in the study is the 23, 265 images of rice leaf in 10 different classes including healthy and diseased. To (hopefully) bypass the shortcomings of single-architecture models, the hybrid model proposed replaces local feature extraction with EfficientNetB0 and global contextual learning with TinyViT as well as a lightweight version of MobileNet. The experimental findings indicate that the hybrid method is better than single CNN and independent Transformer models as it has an accuracy of 98.04%with high precision, recall, and F1-score and low costs of computation. The results demonstrate the usefulness of lightweight CNN-Transformer models in real-time and resource-limited agricultural disease prediction and farm health crop management.", "source_url": "https:\/\/openalex.org\/W7202234903", "pdf_url": "https:\/\/dspace.bracu.ac.bd\/bitstreams\/270bad3b-32c1-4c00-9d85-93d3776dcfb2\/download", "doi": "", "handle": "", "repository_name": "BRAC University Institutional Repository (BRAC University)", "source_record_id": "W7202234903", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 476, "title": "Disease detection and mapping in olive groves using UAVs and deep learning for precision agriculture", "title_ar": "", "title_en": "Disease detection and mapping in olive groves using UAVs and deep learning for precision agriculture", "author": "Maurício Morais", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Portugal", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "object-detection", "categories": "deep-learning; computer-vision; medical-ai; robotics; ai-applications; object-detection; ai-agriculture", "keywords": "", "abstract": "This dissertation presents the implementation and validation of a cost-effective, Unmanned Aerial Vehicle (UAV) based system for automated detection and spatial mapping of olive knot disease in olive groves. Addressing the need for accessible and efficient plant disease monitoring in Precision Agriculture (PA), the proposed methodology leverages existing UAV imagery capabilities with lightweight Deep Learning (DL) models, specifically the You Only Look Once (YOLO) object detection architecture, to enable scalable and accurate detection. The academic contributions presented in this work have resulted in two peer-reviewed publications related to the dissertation topic, as detailed at the end of this document. An annotated dataset of UAV-acquired images was compiled, and several state-of-the-art YOLO object detection models were trained and evaluated under identical conditions. The best-performing model achieved a strong F1-score, demonstrating good results in detecting olive knot disease and accurately mapping its spatial distribution within the plantation. The workflow integrates spatial cross-referencing of detections with UAV flight path data and proximity analysis, enabling the assignment of disease detections to individual trees. An interactive map interface, developed using the Folium Python library, provides visualization of the disease distribution and supports practical grove management. The experimental results indicate that cost-effective UAVs and lightweight DL models can be effectively combined for plant disease detection and spatial analysis, offering a robust and scalable approach for real-world agricultural applications. Limitations regarding early symptom detection and image quality are discussed, and directions for future work are proposed.", "source_url": "https:\/\/openalex.org\/W7112960058", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT)", "source_record_id": "W7112960058", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 477, "title": "A Unified Framework for Distributed and Resource-Aware Plant Disease Detection Using Federated and Adaptive Hybrid Deep Learning", "title_ar": "", "title_en": "A Unified Framework for Distributed and Resource-Aware Plant Disease Detection Using Federated and Adaptive Hybrid Deep Learning", "author": "Ticauris S Stokes", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Southern Mississippi", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "federated-learning", "categories": "machine-learning; deep-learning; medical-ai; robotics; responsible-ai; ai-applications; predictive-analytics; federated-learning; fairness; ai-agriculture; ai-smart-cities", "keywords": "", "abstract": "Plant disease detection in precision agriculture must operate under significant computational constraints, particularly in low-resource (LR) environments such as unmanned aerial vehicles (UAVs) and edge devices. While LR models enable efficient real-time inference, they often lack the representational capacity required to accurately detect complex disease patterns. In contrast, high-resource (HR) models achieve superior detection performance due to their deep architectures but are computationally expensive and unsuitable for direct deployment in resource-constrained settings. This fundamental trade-off between efficiency and accuracy presents a major challenge for practical agricultural monitoring systems. To address the distributed nature of agricultural data, federated learning (FL) is adopted to enable collaborative model training across multiple data sources without requiring centralized data aggregation. FL improves generalization and preserves data privacy; however, it does not resolve the inherent computational disparity between LR and HR models. Consequently, models trained under FL still face limitations when deployed in real-time LR environments. To bridge this gap, this work proposes a hybrid adaptive inference framework that integrates LR and HR models within a unified system. The framework dynamically selects between efficient LR inference and more accurate HR processing based on prediction confidence, enabling effective utilization of computational resources while maintaining high detection performance. Furthermore, a Bayesian optimization strategy is employed to determine optimal decision thresholds, ensuring a balanced trade-off between accuracy, processing time, and energy consumption. Experimental results demonstrate that the proposed framework significantly reduces computational cost while preserving high detection accuracy, enabling scalable and efficient plant disease detection in real-world UAV-based agricultural systems.", "source_url": "https:\/\/openalex.org\/W7162049655", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Aquila Digital Community (University of Southern Mississippi)", "source_record_id": "W7162049655", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 478, "title": "Transfer Learning-Based EfficientNetB0 Model for Multi-Crop Plant Disease Classification", "title_ar": "", "title_en": "Transfer Learning-Based EfficientNetB0 Model for Multi-Crop Plant Disease Classification", "author": "Mebrahtu Yohannes", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Ethiopia", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "cnn", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; ai-applications; transfer-learning; cnn; ai-agriculture", "keywords": "", "abstract": "globally. Early and accurate diagnosis of diseases is key to reducing losses and improving agricultural practices. Traditional methods of plant disease diagnosis are known for being lengthy, cumbersome, and prone to human error. Recent developments in deep learning have shown promise in the automatic diagnosis and classification of plant diseases. Namely, convolutional neural networks (CNNs) perform quite well for such problems. The purpose of this study is to present an approach to plant disease classification that is based on the use of transfer learning and utilizes the Convolutional Neural Network architecture called EfficientNetB0. Transfer learning with ImageNet pretrained weights and an efficient architecture for feature extraction was chosen because of its ability to significantly lower the computational resources required and improve performance. Moreover, plant leaf images from different datasets were taken into account to provide better training. The model achieved 99.32 % training accuracy, 99.37% validation accuracy, and 99.32% test accuracy. It also had impressive convergence capability and low computational complexity, with around 4.7 million parameters and the size of the saved model at around 34 MB. It can be concluded from the findings that the transfer learning-based EfficientNetB0 model provides a great balance between accurate classification and efficiency, while improving the practicality of the system.", "source_url": "https:\/\/openalex.org\/W7212441753", "pdf_url": "", "doi": "10.82589\/muir-1230", "handle": "", "repository_name": "Mekelle University Institutional Repository (MU-IR)", "source_record_id": "W7212441753", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 480, "title": "Plant disease classification system using deep learning", "title_ar": "", "title_en": "Plant disease classification system using deep learning", "author": "Mohd Azizul Asmad Latip", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Universiti Malaysia Sabah", "university_source": "repository_host", "college": "", "department": "", "country": "Malaysia", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "medical-ai", "research_subdomain": "cnn", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; ai-applications; cnn; ai-cybersecurity; ai-agriculture", "keywords": "", "abstract": "In agriculture production, the unlimited and no disease plant product from farming become important things due this product is necessities in terms foods and mainly source for highly nutrient to community regardless country in the world. However, an increase in the human population requires an increase in agricultural production. Generally, the most important thing in agriculture that affects the quantity and quality of crops is plant diseases and make crop disease become a major threat to food security. In general, a farmer knows that his plant is attacked by a disease through direct vision. But, this process is sometimes inaccurate. Plant diseases are not only a threat to food security at the global scale, but can also have disastrous consequences for smallholder farmers whose livelihoods depend on healthy crops. With the development of machine learning technology, plant disease classification can be done automatically using deep learning. In deep learning, a convolutional neural network (CNN) is a class of deep neural networks, most commonly applied to analysing visual imaginary. They are also known as shift invariant or space invariant artificial neural networks (SIANN), based on the shared-weight architecture of the convolution kernels that scan the hidden layers and translation invariance characteristics.", "source_url": "https:\/\/openalex.org\/W7198848850", "pdf_url": "https:\/\/eprints.ums.edu.my\/id\/eprint\/33302\/1\/PLANT%20DISEASE%20CLASSIFICATION%20SYSTEM%20USING%20DEEP%20LEARNING.24pages.pdf", "doi": "", "handle": "", "repository_name": "UMS Institutional Repository (Universiti Malaysia Sabah)", "source_record_id": "W7198848850", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 485, "title": "Deep Learning for Plant Disease Detection and Classification on Field Images", "title_ar": "", "title_en": "", "author": "Emmanuel Moupojou Matango", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2024, "language": "fr", "research_domain": "medical-ai", "research_subdomain": "image-segmentation", "categories": "deep-learning; computer-vision; medical-ai; ai-applications; image-segmentation; ai-agriculture", "keywords": "", "abstract": "Apprentissage profond pour la détection des maladies de plantes sur des images de terrain Plusieurs modèles d’apprentissage profond ont été proposées pour aider les agriculteurs à détecter le plus efficacement possible les maladies des cultures et éviter les pertes de rendement. Ces modèles sont généralement entrainés sur des bases de données, telles que PlantVillage, composées d'images de plantes prises dans des conditions de laboratoire avec des feuilles uniques et un arrière-plan uniforme. Un réseau de neurones entraîné sur une telle base de données a une très faible précision lorsqu'il est exécuté sur des images de terrain qui ont quant à elles des arrière-plans complexes et plusieurs feuilles par image. Dans cette thèse, nous proposons premièrement la base de données FieldPlant contenant 5 170 images de maladies des plantes collectées dans les plantations et annotées manuellement. Deuxièmement, nous proposons une combinaison de modèles pour identifier avec précision les maladies des plantes sur des images de terrain. Cette solution utilise le modèle de segmentation universel pour extraire efficacement tous les objets identifiables dans une image. Pour distinguer les objets d'arrière-plan identifiés et les objets feuilles, nous utilisons la Description de Données Entièrement Convolutives, un modèle de classification profond à une classe pour la détection d'anomalies. Notre modèle peut détecter plusieurs maladies apparaissant sur différentes feuilles dans la même image et améliore de 15% la précision de classification sur des bases de données publiques de maladies de plantes telles que PlantDoc.", "source_url": "https:\/\/openalex.org\/W7152640065", "pdf_url": "http:\/\/www.theses.fr\/2024TROY0008\/document", "doi": "10.70675\/6c91b081z9504z442dz842fzab0592a5e598", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7152640065", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 491, "title": "Plant Species Detection in Aerial and Satellite Images using Deep Learning", "title_ar": "", "title_en": "Plant Species Detection in Aerial and Satellite Images using Deep Learning", "author": "Anastasiia N. Safonova", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Granada", "university_source": "repository_host", "college": "", "department": "", "country": "Spain", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "computer-vision", "research_subdomain": "", "categories": "deep-learning; computer-vision; robotics", "keywords": "", "abstract": "It is unquestionable that the conservation of plant species is essential for the preservation of nature, climate, and human wellbeing.Classically, the task of conserving threatened plant species was generally done through direct field supervision by natural resources managers.Thanks to the advances in sensor technologies and also in aircrafts such as unmanned aerial vehicles (UAV), high resolution remote sensing (RS) data are becoming an important resource for monitoring threatened plant species in large areas.Such RS data are usually multi-spectral images, with three or more bands, of up to 3 cm\/pixel resolution, providing an orthogonal or quasi-orthogonal view of the considered plant species.This information may not be as complete as the information provided by natural images; however, it might be sufficient to monitor tree species located in very large areas with difficult access.", "source_url": "https:\/\/openalex.org\/W3183821534", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Institutional Repository of the University of Granada (University of Granada)", "source_record_id": "W3183821534", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 501, "title": "Quantization on Graph Neural Networks for Image Classification", "title_ar": "", "title_en": "Quantization on Graph Neural Networks for Image Classification", "author": "Rithik Reddy Katpally", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "San Jose State University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "computer-vision", "research_subdomain": "image-classification", "categories": "machine-learning; deep-learning; computer-vision; transfer-learning; cnn; gnn; image-classification", "keywords": "", "abstract": "Quantization has become a key approach for reducing storage and computational demands of deep neural networks while maintaining high accuracy. Although 8-bit quantization is well-established for convolutional architectures such as ResNet50 and MobileNetV2, its application to graph-based vision models remains underexplored. In this work, we extend quantization-aware training to Vision Graph Neural Networks (ViGs) and conduct comparisons with quantized CNNs on the CIFAR-100 dataset. To ensure parity, all models have same training hyperparameters such as learning rate, batch size, optimizer, number of epochs. We used numerous techniques to preserve performance for low-bit precision. First, Pauta Quantization clips activation outliers based on statistical thresholds. Next, Attention Quantization Distillation (AQD) encourages the quantized network to mimic channel-wise attention patterns from full-precision activations to retain feature representations. Finally, Stochastic Quantization Distillation (SQD) injects randomness into the quantization process to make it more robust. Comprehensive experiments on CIFAR-100 reveal that 8-bit quantization delivers an excellent trade-off between efficiency and accuracy across both architectures. Notably, our quantized ViGs almost matches the performance of their ResNet50 and MobileNetV2 for top-1 accuracy, while also making similar inference and memory footprint. These findings demonstrate that quantized Vision GNNs are a practical alternative to CNNs for deployment on resource-constrained devices. Future research will explore fine-tuning of AQD and other distillation parameters.", "source_url": "https:\/\/openalex.org\/W4413638221", "pdf_url": "", "doi": "10.31979\/etd.d5rw-djyy", "handle": "", "repository_name": "San José State University ScholarWorks (San Jose State University)", "source_record_id": "W4413638221", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 504, "title": "Extending Graph Neural Networks with Global Features", "title_ar": "", "title_en": "", "author": "Dragos-Andrei Brasoveanu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "TU Wien", "university_source": "repository_host", "college": "", "department": "", "country": "Austria", "governorate": "", "publication_year": 2026, "language": "de", "research_domain": "deep-learning", "research_subdomain": "gnn", "categories": "deep-learning; gnn", "keywords": "", "abstract": "Wir untersuchen die Verwendung globaler Graphmerkmale, um die Ausdruckskraft und die Vorhersageleistung von graph neural networks (GNNs) zu verbessern. Während Message Passing Neural Networks (MPNNs) ein gängiger Ansatz für das Lernen von Graphrepräsentationen sind, sind sie in ihrer Ausdruckskraft beschränkt. Diese Einschränkung beeinträchtigt die Universalität von MPNNs und folglich die Arten von Grapheneigenschaften, die sie darstellen können.Um diese Herausforderungen anzugehen, schlagen wir vor, expressive globale Graphmerkmale in eine Basis-GNN-Architektur zu integrieren. Im Gegensatz zu herkömmlichen Ansätzen, die die Vorhersageleistung entweder durch Änderungen im Message Passing oder durch Modifikation der Graphen (z. B. Hinzufügen zusätzlicher Knotenmerkmale) verbessern, konzentriert sich unsere Methode auf globale Graphmerkmale, die die gesamte Graphstruktur beschreiben. Diese globalen Merkmale, die in Bereichen wie der Chemoinformatik gut etabliert, jedoch in der GNN Literatur oft übersehen sind, können verwendet werden, um wichtige Aspekte des Graphen zu erfassen, die von MPNNs nicht ausgedrückt werden können, wie etwa den Graphendurchmesser oder den längsten Zyklus.Wir ergänzen die von GNNs gelernten Embeddings mit ausgewählten globalen Graphmerkmalen und deren Einfluss auf die Ausdruckskraft und Modellleistung. Wir zeigen, dass 19 der 23 untersuchten globalen Merkmale, wie der Wiener- oder der Hosoya-Index, nachweislich die Ausdruckskraft von MPNNs erhöhen. Unsere empirischen Studien belegen, dass die Integration globaler Graphmerkmale die Vorhersagekraft von GNNs bei molekularen Benchmark-Datensätzen verbessern kann, was auf eine vielversprechende Richtung zur Steigerung der Effektivität von GNN-basierten Modellen hinweist.", "source_url": "https:\/\/openalex.org\/W7138864784", "pdf_url": "https:\/\/repositum.tuwien.at\/bitstream\/20.500.12708\/227048\/1\/Brasoveanu%20Dragos-Andrei%20-%202026%20-%20Extending%20Graph%20Neural%20Networks%20with%20Global...pdf", "doi": "10.34726\/hss.2026.120129", "handle": "", "repository_name": "reposiTUm (TU Wien)", "source_record_id": "W7138864784", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 505, "title": "Descriptive Complexity and Graph Neural Networks", "title_ar": "", "title_en": "", "author": "Matias Selin", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Tampere University", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2025, "language": "fi", "research_domain": "deep-learning", "research_subdomain": "gnn", "categories": "deep-learning; gnn", "keywords": "", "abstract": "Koneoppimisen sisällä kehittynyt syväoppiminen on erityisesti viimeisen vuosikymmenen aikana noussut yhdeksi merkittävimmistä tämänhetkisistä tutkimusaiheista [68]. Suuri osa kehityksestä on kuitenkin perustunut kroonisen epäluotettavaan empiiriseen tutkimukseen samalla kun aiheen matemaattinen teoria on jäänyt pienemmälle huomiolle [40]. Erityistapauksessa, jossa käsiteltävänä on graafimuotoinen data, ovat graafineuroverkot [33, 31] saavuttaneet merkittävän suosion huolimatta siitä, että niiden teoreettiset perusominaisuudet ovat pitkään jääneet epämääräisiksi. Esimerkiksi näiden mallien ilmaisuvoimaa, eli niiden kykyä ilmaista eri funktioita syötetyllä datalla, on vasta hiljattain ymmärretty paremmin [11]. Yhdeksi lupaavaksi lähestymistavaksi tässä on osoittautunut kuvaileva vaativuusteoria, jossa tavoitteena on löytää näitä malleja karakterisoivia logiikoita [41]. Tässä tutkielmassa käsittelen kuvailevan vaativuusteorian soveltamista graafineuroverkkoihin. Luvun 1 johdannon jälkeen aloitan tutkielman varsinaisen sisällön luvussa 2 kattavalla kirjallisuuskatsauksella aiheen historiasta, minkä jälkeen kuvailen yksityiskohtaisesti luvussa 3 miten Ahvosen ja muiden [8] rekurrenttien liukuluvuilla laskevien graafineuroverkkojen karakterisaatio laskevalla modaalisubstituutiokalkyylillä toimii. Kuvailen myös, miten globaalin lukemisen lisääminen graafineuroverkkoihin voidaan simuloida logiikan puolella tätä vastaavalla globaalilla laskevalla modaalitimantilla. Luvussa 4 tarkastelen erästä tunnettua graafineuroverkkojen erotteluvoiman rajaa eli Weisfeilerin ja Lemanin algoritmia [86, 61] sekä sitä, millä keinoin sen yli voidaan päästä. Eri vaihtoehdoista tehokkain on Saton ja muiden [77] kehittämä noodirandomointi, jossa – hieman yksinkertaistettuna – jokaiselle noodille asetetaan yksilöllinen tunniste. Osoitan uutena tuloksena, että Ahvosen ja muiden [8] karakterisaatiotulosta voidaan helposti laajentaa noodirandomoituun tapaukseen sallimalla logiikassa propositiosymbolien kvantifiointi. Tutkielman lopuksi tutkin tätä uutta logiikkaa tarkemmin muun muassa vertailemalla sitä erääseen sen muunnokseen, joka poikkeaa siitä syntaktisesti vain hieman mutta osoittautuu muilta ominaisuuksiltaan varsin erilaiseksi. Tämän luvun tulokset perustuvat yhteistyöhön Antti Kuusiston kanssa.", "source_url": "https:\/\/openalex.org\/W7133053626", "pdf_url": "https:\/\/trepo.tuni.fi\/bitstream\/10024\/232168\/2\/SelinMatias.pdf", "doi": "", "handle": "", "repository_name": "Tampere University Institutional Repository (Tampere University)", "source_record_id": "W7133053626", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 509, "title": "Anomaly Detection Using Graph Neural Network", "title_ar": "", "title_en": "Anomaly Detection Using Graph Neural Network", "author": "Bishal Lakha", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Boise State University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; ai-applications; transformers; gnn; object-detection; ai-cybersecurity", "keywords": "", "abstract": "Detecting malicious behavior is becoming increasingly crucial as the internet becomes more prevalent. This problem can be formulated as an anomaly detection task on provenance data, where attacks are detectable as anomalies in the behavior of the system. The availability of system-level data in comparison to network data is quite limited and so is the research carried out on system-level logs. However, monitoring the operating system's processes during program execution and identifying anomalous behavior in system calls can be beneficial since it can provide broad coverage and generality, as a variety of malicious applications could be identified. Furthermore, logs like system processes and events are provenance data- a graph that describes the relationship between all the elements that contributed to the creation of the data, making use of a Graph Neural Network (GNN) better suited for the task. Moreover, such data may contain metadata, which in general tends to be complex and make feature engineering more difficult resulting in limited usage of such features. In this thesis, we address these issues by first utilizing the graph-like structure of logs, in which processes enact events and generate additional processes. Then we use a graph neural network to create representations of each event, encoding information about their neighboring events in a way that is unsupervised. The second is to make use of complex features such as command arguments which vary widely and cannot be used in the presented format as features in typical machine learning algorithms. If these features are instead encoded using a system composed of transformer and Variational Auto Encoder models, they can then be used in other algorithms such as a GNN or anomaly detector. These two approaches combined improve anomaly detection AUCROC for the BETH dataset by around 8 percent as compared to the manually engineered features alone.", "source_url": "https:\/\/openalex.org\/W4391694518", "pdf_url": "", "doi": "10.18122\/td.2065.boisestate", "handle": "", "repository_name": "ScholarWorks (Boise State University)", "source_record_id": "W4391694518", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 512, "title": "Analyzing and improving graph neural networks", "title_ar": "", "title_en": "", "author": "Guillaume Renton", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2021, "language": "fr", "research_domain": "computer-vision", "research_subdomain": "gnn", "categories": "deep-learning; computer-vision; gnn", "keywords": "", "abstract": "Réseaux de neurones sur graphes : analyse et contributions Bien que théorisés il y a une quinzaine d’années, l’intérêt de la communauté scientifique pour les réseaux de neurones sur graphes n’a connu un réel essor que très récemment. De tels modèles visent à transposer les capacités d’apprentissage de représentation inhérentes aux réseaux de neurones profonds sur des données de type graphes, via l’apprentissage d’états cachés associés aux nœuds du graphe. Ces états cachés sont calculés et mis à jour en fonction des informations contenues dans le voisinage de chacun des nœuds. Ce récent intérêt pour les réseaux de neurones sur graphes (GNN) a conduit à une \"jungle\" de modèles et de méthodes, rendant le domaine de recherche parfois confus. Historiquement, deux principales stratégies ont été explorées : les réseaux spatiaux et les réseaux spectraux. Les réseaux spatiaux, parfois appelés \"message passing neural network\", sont basés sur le calcul d’un message agrégeant l’information contenue dans le voisinage de chacun des nœuds. Ce message est ensuite utilisé afin de mettre à jour les états cachés des différents nœuds du graphe. Les réseaux spectraux quant à eux sont basés sur la théorie spectrale des graphes et reposent donc sur le Laplacien du graphe. La décomposition en valeurs\/vecteurs propres du Laplacien permet notamment de définir une transformée de Fourier sur graphe ainsi qu’une transformée inverse. À partir de ces transformées, différents filtrages peuvent être appliqués sur le graphe, obtenant des résultats similaires au filtrage sur une image ou sur un signal. Dans cette thèse, nous commençons par introduire une troisième catégorie, appelée spectral-rooted spatial convolution. En effet, certaines méthodes récentes prennent racine dans le domaine spectral tout en évitant le calcul de la décomposition en vecteurs propres du Laplacien. Cette troisième catégorie nous amène à nous poser la question de la différence fondamentale entre les réseaux de neurones spatiaux et réseaux de neurones spectraux. Nous répondons à cette question par la proposition d’un modèle général unifiant les deux stratégies, montrant notamment que les modèles spectraux sont un cas particulier des modèles spatiaux. Ce cadre unifié nous a par ailleurs permis de proposer une analyse spectrale de plusieurs modèles de GNN populaires dans la communauté scientifique, à savoir GCN, GIN, GAT, Chebnet et CayleyNet. Cette analyse montre que les modèles spatiaux sont limités aux filtrages passe-bas et passe haut, tandis que les modèles spectraux sont capables de réaliser n’importe quel type de filtres. Ces résultats sont ensuite retrouvés avec la présentation d’un problème jouet, montrant dans un premier temps la limitation des modèles spatiaux à définir des filtres passe-bande et l’importance que peut revêtir l’utilisation de tels filtres.Ces résultats nous ont amenés à proposer une méthode capable de réaliser n’importe quel type de filtrage, tout en limitant le nombre de paramètres du réseau. En effet, bien que les modèles spectraux soient capables de réaliser tout type de filtrage, l’ajout d’un nouveau filtre requiert l’ajout d’une nouvelle matrice de poids dans le réseau de neurones. Afin de réduire le nombre de paramètres, nous proposons l’adaptation des Depthwise Separable Convolution aux graphes via une méthode intitulée Depthwise Separable Graph Convolution Network. Cette méthode a été évaluée à la fois en apprentissage transductif et en apprentissage inductif, et obtient des résultats supérieurs à l’état de l’art sur les datasets testés. Enfin, nous proposons une méthode définie dans le domaine spatial afin de prendre en compte les attributs d’arcs. En effet, cette problématique a été peu étudiée par la communauté scientifique, et le nombre de méthodes proposant d’inclure ces informations est très réduit.", "source_url": "https:\/\/openalex.org\/W7151004721", "pdf_url": "https:\/\/theses.hal.science\/tel-03346018v1\/document", "doi": "10.70675\/e5978f02z5c61z4768zb855z5c9574d2cea4", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7151004721", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 513, "title": "Track reconstruction using graph neural networks", "title_ar": "", "title_en": "Track reconstruction using graph neural networks", "author": "Cruz Carpio, Carlos Andrés", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Cantabria", "university_source": "repository_host", "college": "", "department": "", "country": "Spain", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "deep-learning; computer-vision; gnn; object-detection", "keywords": "", "abstract": "After more than a decade of data-taking of the experiments at the Large Hadron Collider at CERN, no evidence of New Physics has been found. A new accelerator, the High-Luminosity Large Hadron Collider is currently under construction, increasing the instantaneous luminosity by one order of magnitud. The new accelerator conditions generate a challenging environment for the detectors at the LHC. One of the most important difficulties that the CMS experiment will find is the extremely large hit occupancy in the tracker and the huge combinatorial problem associated to the track reconstruction. This is even more difficult for displaced topologies emerging from potential new, long-lived particles, where the traditional seeding methods cannot be used. This work explores the use of Graph Neural Networks as an alternative procedure to identify tracks in a detector. A simple track generator has been developed together with a simplified model of the CMS tracker, and the resulting tracks and hits are reconstructed using a dedicated GNN algorithm trained with a number of example tracks. The performance of the algorithm is tested to be higher than 80% even for 500 simultaneous tracks.", "source_url": "https:\/\/openalex.org\/W7110671432", "pdf_url": "https:\/\/repositorio.unican.es\/xmlui\/bitstream\/10902\/35392\/1\/CruzCarpioCarlosAndres.pdf", "doi": "", "handle": "", "repository_name": "UCrea (University of Cantabria)", "source_record_id": "W7110671432", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 519, "title": "Power Flow Optimization with Graph Neural Networks", "title_ar": "", "title_en": "Power Flow Optimization with Graph Neural Networks", "author": "Jonas Berg Hansen", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Norway", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "deep-learning", "research_subdomain": "gnn", "categories": "deep-learning; ai-applications; gnn; ai-smart-cities", "keywords": "", "abstract": "Power flow analysis is an important tool in power engineering for planning and operating power systems. The standard power flow problem consists of a set of non-linear equations, which are traditionally solved using numerical optimization techniques, such as the Newton-Raphson method. However, these methods can become computationally expensive for larger systems, and convergence to the global optimum is usually not guaranteed. In recent years, several methods using Graph Neural Networks (GNNs) have been proposed to speed up the computation of the power flow solutions, without making large sacrifices in terms of accuracy. This class of models can learn localized features that are independent from a global graph structure. Therefore, by representing power systems as graphs these methods can, in principle, generalize to systems of different size and topology. However, most of the current approaches have only been applied to systems with a fixed topology and none of them were trained simultaneously on systems of different topology. Hence, these models are not fully shown to generalize to widely different systems or even to small perturbations of a given system. In this thesis, several supervised GNN models are proposed to solve the power flow problem, using established GNN blocks from the literature. These GNNs are trained on a set of different tasks, where the goal is to study the generalizability to both perturbations and completely different systems, as well as comparing performance to standard Multi-Layered Perceptron (MLP) models. The experimental results show that the GNNs are comparatively successful at generalizing to widely different topologies seen during training, but do not manage to generalize to unseen topologies and are not able to outperform an MLP on slight perturbations of the same energy system. The study presented in this thesis allowed to draw important insights about the applicability of GNN as power flow solvers. In the conclusion, several possible ways for improving the GNN-based solvers are discussed.", "source_url": "https:\/\/openalex.org\/W3200352710", "pdf_url": "", "doi": "", "handle": "", "repository_name": "NORA - Norwegian Open Research Archives", "source_record_id": "W3200352710", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 528, "title": "Decision tree learning for intelligent mobile robot navigation", "title_ar": "", "title_en": "Decision tree learning for intelligent mobile robot navigation", "author": "G.H. Shah Hamzei", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Loughborough University", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "generative-ai", "research_subdomain": "ai-smart-cities", "categories": "generative-ai; robotics; expert-systems; ai-applications; ai-smart-cities", "keywords": "", "abstract": "The replication of human intelligence, learning and reasoning by means of computer algorithms is termed Artificial Intelligence (Al) and the interaction of such algorithms with the physical world can be achieved using robotics. The work described in this thesis investigates the applications of concept learning (an approach which takes its inspiration from biological motivations and from survival instincts in particular) to robot control and path planning. The methodology of concept learning has been applied using learning decision trees (DTs) which induce domain knowledge from a finite set of training vectors which in turn describe systematically a physical entity and are used to train a robot to learn new concepts and to adapt its behaviour. To achieve behaviour learning, this work introduces the novel approach of hierarchical learning and knowledge decomposition to the frame of the reactive robot architecture. Following the analogy with survival instincts, the robot is first taught how to survive in very simple and homogeneous environments, namely a world without any disturbances or any kind of \"hostility\". Once this simple behaviour, named a primitive, has been established, the robot is trained to adapt new knowledge to cope with increasingly complex environments by adding further worlds to its existing knowledge. The repertoire of the robot behaviours in the form of symbolic knowledge is retained in a hierarchy of clustered decision trees (DTs) accommodating a number of primitives. To classify robot perceptions, control rules are synthesised using symbolic knowledge derived from searching the hierarchy of DTs. A second novel concept is introduced, namely that of multi-dimensional fuzzy associative memories (MDFAMs). These are clustered fuzzy decision trees (FDTs) which are trained locally and accommodate specific perceptual knowledge. Fuzzy logic is incorporated to deal with inherent noise in sensory data and to merge conflicting behaviours of the DTs. In this thesis, the feasibility of the developed techniques is illustrated in the robot applications, their benefits and drawbacks are discussed.", "source_url": "https:\/\/openalex.org\/W7204433763", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Loughborough University Research Repository (Loughborough University)", "source_record_id": "W7204433763", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 529, "title": "Knowledge Representation and Automated Task Planning for Intelligent Robots through Spatial Temporal Logic and Learning", "title_ar": "", "title_en": "Knowledge Representation and Automated Task Planning for Intelligent Robots through Spatial Temporal Logic and Learning", "author": "Zhiyu Liu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "video-analysis", "categories": "machine-learning; computer-vision; generative-ai; robotics; video-analysis", "keywords": "", "abstract": "Intelligent robots have emerged as a hot research topic in the interdisciplinary study of control theory, robotics, and computer science as it demonstrates huge potential in fundamentally benefiting our society. The development of intelligent robots with the ability to independently solve a new task is an important yet challenging research problem. This work devotes to proposing an integrated system for such intelligent robots through three major objectives. First, we propose a new spatial temporal logic with a sound and complete deduction system for knowledge representation and reasoning in intelligent robots. The proposed logic achieves a balance between expressiveness and tractability for intelligent robots since the satisfiability of the proposed logic is decidable. We propose two specification mining algorithms to generate spatial temporal logic formulas from video and unstructured text data. The obtained formulas are used to construct a domain theory storing necessary knowledge for task planning. Second, we develop automatic task planning with an interactive proposer and verifier for intelligent robots based on the domain theory in spatial temporal logic. The proposer is used to generate ordered actions through forward graph expansion and backward solution extraction, and the verifier is used to verify the results from the proposer and generate executable task plans. We further study the scenario where the domain theory for task planning is incomplete by integrating the task planning algorithm with machine learning systems. The proposed framework gives intelligent robots the ability to actively seeking knowledge online based on missing information. Third, we study a motion planning problem for intelligent robots under formal specifications of the proposed spatial temporal logic. We develop a control synthesis algorithm for the formal specifications by proposing a two-layer MPC framework. The high-layer transfers the global formal specifications into a sequence of sub-formulas and the low-layer solves a distributed MPC using an MILP solver. We show that the proposed framework is sound and recursively feasible.", "source_url": "https:\/\/openalex.org\/W3202693723", "pdf_url": "", "doi": "10.7274\/1831cj85f59", "handle": "", "repository_name": "Figshare", "source_record_id": "W3202693723", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 530, "title": "Mobile Robot Learning", "title_ar": "", "title_en": "Mobile Robot Learning", "author": "Gregory Kahn", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "predictive-analytics", "categories": "machine-learning; computer-vision; reinforcement-learning; robotics; predictive-analytics", "keywords": "", "abstract": "In order to create mobile robots that can autonomously navigate real-world environments, we need generalizable perception and control systems that can reason about the outcomes of navigational decisions. Learning-based methods, in which the robot learns to navigate by observing the outcomes of navigational decisions in the real world, offer considerable promise for obtaining these intelligent navigation systems. However, there are many challenges impeding mobile robots from autonomously learning to act in the real-world, in particular (1) sample-efficiency---how to learn using a limited amount of data? (2) supervision---how to tell the robot what to do? and (3) safety---how to ensure the robot and environment are not damaged or destroyed during learning?In this thesis, we will present deep reinforcement learning methods for addressing these real world mobile robot learning challenges. At the core of these methods is a predictive model, which takes as input the current robot sensors and predicts future navigational outcomes; this predictive model can then be used for planning and control. We will show how this framework can address the challenges of sample-efficiency, supervision, and safety to enable ground and aerial robots to navigate in complex indoor and outdoor environments.", "source_url": "https:\/\/openalex.org\/W3161107037", "pdf_url": "", "doi": "", "handle": "", "repository_name": "eScholarship (California Digital Library)", "source_record_id": "W3161107037", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 533, "title": "Design and development of an intelligent mobility robot", "title_ar": "", "title_en": "Design and development of an intelligent mobility robot", "author": "Kun Huang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "deep-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; deep-learning; robotics; expert-systems; ai-applications; predictive-analytics; ai-industry; ai-smart-cities", "keywords": "", "abstract": "To meet the demand of short-distance transportation, a novel type of intelligent mobility robot is proposed by this thesis. The robot features high portability, reliable safety and multiple control and interaction methods, including posture control and remote interaction. The robot has wide range of application prospects in both industrial production and daily life, such as travel, services, entertainment, etc. To realize posture control of the robot, a complete set of posture perception system is designed and developed. Firstly, the pressures on different sections are real-time acquired by specific sensors and corresponding detection circuit. Then based on the pressure distributions and statistic information, the posture is perceived by a series of processing algorithms, including sampling a reference posture when the user is standing stably and calculating current user posture by an non-linear mapping. Next, the posture is quantitatively described as the inclination angles of the body gravity center in two dimensions. Moreover, two machine learning are applied to further improve the accuracy of posture detection, including parameter regression method and Gaussian Process Dynamic Model. In addition, some auxiliary sensors are also applied to ensure the robustness and reliability through fusion of multi-sensor information. At last, with the help of hybrid neural network, some interaction via posture patterns are achieved. This thesis also analyzes the planar kinematics of the intelligent mobility robot, considering its steering mechanisms, suspension systems, and differential speed of wheel pairs. The motion planning algorithm outputs desired motion parameters of the robot, such as longitudinal speed and turning curvature, based on both user input and kinematic constrains. To control the motion of the robot, the distribute control strategy as well as fuzzy logic control algorithm are applied to coordinately assign appropriate control amount to multiple actuators of different types. The design and control methods of actuators are discussed at last. This thesis designs and implements the Field Oriented Control to drive the Permanent Magnet Synchronous Motor and to control its speed and toque. The FOC system improves energy efficiency and decreases the operation noise and toque ripple. Meanwhile, a kind of brake device is also designed by the thesis to provide reliable and timely braking performance, wherein the braking force is adjustable.", "source_url": "https:\/\/openalex.org\/W2939709170", "pdf_url": "", "doi": "10.14711\/thesis-991012554966503412", "handle": "", "repository_name": "", "source_record_id": "W2939709170", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 536, "title": "Socially intelligent robots that understand and respond to human touch", "title_ar": "", "title_en": "Socially intelligent robots that understand and respond to human touch", "author": "Merel M. Jung", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "medical-ai", "research_subdomain": "", "categories": "machine-learning; medical-ai; robotics", "keywords": "", "abstract": "Touch is an important nonverbal form of interpersonal interaction which is used to communicate emotions and other social messages. As interactions with social robots are likely to become more common in the near future these robots should also be able to engage in tactile interaction with humans. Therefore, the aim of the research presented in this dissertation is to work towards socially intelligent robots that can understand and respond to human touch. To become a socially intelligent actor a robot must be able to sense, classify and interpret human touch and respond to this in an appropriate manner. To this end we present work that addresses different parts of this interaction cycle. The contributions of this dissertation are the following. We have made a touch gesture dataset available to the research community and have presented benchmark results. Furthermore, we have sparked interest into the new field of social touch recognition by organizing a machine learning challenge and have pinpointed directions for further research. Also, we have exposed potential difficulties for the recognition of social touch in more naturalistic settings. Moreover, the findings presented in this dissertation can help to inform the design of a behavioral model for robot pet companions that can understand and respond to human touch. Additionally, we have focused on the requirements for tactile interaction with robot pets for health care applications.", "source_url": "https:\/\/openalex.org\/W2744115739", "pdf_url": "https:\/\/ris.utwente.nl\/ws\/files\/12836170\/Dissertation_Merel_Jung.pdf", "doi": "10.3990\/1.9789036543644", "handle": "", "repository_name": "", "source_record_id": "W2744115739", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 537, "title": "Skill Learning for Industrial Robot Manipulators", "title_ar": "", "title_en": "Skill Learning for Industrial Robot Manipulators", "author": "Te Tang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "machine-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; robotics; ai-applications; predictive-analytics; ai-industry", "keywords": "", "abstract": "Industrial robots have been kept upgraded for decades to achieve extraordinary accuracy, speed, and repeatability. However, even the most advanced manipulator today is functioning as a programmable machine, instead of an intelligent agent. This deficiency of intelligence restricts robots from broader applications. To meet the increasing demand for automation, it is essential to make industrial robots more skillful and intelligent. Under this background, the objective of this dissertation is to develop generic and efficient methodologies to teach robots novel skills. Three major approaches including model-based learning, model-free learning and analogy-based learning are discussed and explored. A series of skills such as assembly, grasping, tracking and motion planning, have been successfully taught to industrial manipulators and evaluated by experiments.As the name implies, model-based learning tries to formulate skills analytically based on physical models. In Chapter 2, an auto-alignment skill is developed for robotic assembly by constructing a novel contact model. Robots are enabled to predict tilt angles between assembly parts from force\/torque measurement and perform fine assembly from large misalignment conditions. With this skill model, traditional procedures such as the installation of positioning fixtures and manual alignment can be skipped, which saves tremendous preparation efforts for robotic assembly.However, not always system models can be constructed, especially for those complicated scenarios. The model-free approach is then developed to learn control policies by regressing general parametric functions. In Chapter 3, a compliant robotic force controller is learned from human demonstration. A human operator holds a specially-designed handle and demonstrates to robots the compliant insertions. The Gaussian mixture regression is introduced to fit motion patterns from measured data. This approach enables to transfer the compliant assembly skill from humans to robots efficiently and intuitively. Besides model-based and model-free learning, an analogy-based leaning approach is proposed in Chapter 4. The distinct idea is that instead of pursuing a control policy either constructed by models or regressed from data, we discover the correlation between scenarios, i.e., analogy. The past scenario that bears a strong similarity with the current one will be identified, and a mapping function between these two scenes will be constructed. Applying the mapping function, the past problem-solving action can be transferred to a new one that is feasible for the current scenario. This analogy-based approach has been implemented in multiple industrial tasks, and taught robots various skills such as grasping versatile objects (Chapter 5), tracking and manipulating deformable objects (Chapter 6) and efficient motion re-planning for similar scenarios (Chapter 7).", "source_url": "https:\/\/openalex.org\/W2902534828", "pdf_url": "", "doi": "", "handle": "", "repository_name": "eScholarship (California Digital Library)", "source_record_id": "W2902534828", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 539, "title": "Towards Probabilistic Generative Models for Socially Intelligent Robots", "title_ar": "", "title_en": "Towards Probabilistic Generative Models for Socially Intelligent Robots", "author": "Xavier Alameda-Pineda", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "machine-learning; deep-learning; computer-vision; nlp; generative-ai; robotics; video-analysis; speech-recognition", "keywords": "", "abstract": "Socially intelligent robots represent a scientific and technological challenge that needs significant contributions at the cross-roads of machine learning, computer vision, audio processing and robotics to be addressed. Naturally, given that the communication modalities most commonly used in human interaction are audio and video, audio-visual fusion plays is a skill to be mastered for robots to exhibit social intelligence. The scenarios and tasks to be addressed in social robotics, require learning methods operating in unsupervised settings and handling the uncertainty in perception and in action. In addition, the developed tools cannot disregard the recent advances and breakthrough in representation learning brought by deep neural architectures. Consequently, combinations of probabilistic models – able to account for uncertainty – and deep neural networks – learning powerful data representations – seem the most appropriate methodological framework to develop social intelligence for robotic platforms. In this manuscript, we motivate the interest of developing machine learning methods exploiting audio-visual data to endow robots with social intelligence. We present the foundations of probabilistic generative models and discuss in detail seven recent contributions. Each of them addresses a different task, useful to acquire social intelligence, and are developed under the large framework of deep probabilistic models – combinations of deep neural networks and probabilistic models. Several future research opportunities as well as my personal scientific ambition for the next years conclude the manuscript.", "source_url": "https:\/\/openalex.org\/W4392497979", "pdf_url": "", "doi": "", "handle": "", "repository_name": "HAL", "source_record_id": "W4392497979", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 540, "title": "Robot Learning for Muscular Systems", "title_ar": "", "title_en": "Robot Learning for Muscular Systems", "author": "Dieter Büchler", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Technical University of Darmstadt", "university_source": "repository_host", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "reinforcement-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; reinforcement-learning; robotics; ai-applications; predictive-analytics; ai-agriculture; ai-industry; ai-smart-cities", "keywords": "", "abstract": "Today's robots are capable of performing many tasks that tremendously improve human lives. For instance, in industrial applications, robots move heavy parts very quickly and precisely along a predefined path. Robots are also widely used in agriculture or domestic applications like vacuum cleaning and lawn mowing. However, in more general settings, the gap between human abilities and what current robots deliver is still not bridged, such as in dynamic tasks. Like table tennis with anthropomorphic robot arms, such tasks require the execution of fast motions that potentially harm the system. Optimizing for such fast motions and being able to execute them without impairing the robot still pose difficult challenges that, so far, have not been met. Humans perform dynamic tasks relatively easy at high levels of performance. Can we enable comparable perfection on kinematically anthropomorphic robots? This thesis investigates whether learning approaches on more human-like actuated robots bring the community a step closer towards this ambitious goal. Learning has the potential to alleviate control difficulties arising from fast motions and more complex robots. On the other hand, an essential part of learning is exploration, which forms a natural trade-off with robot safety, especially at dynamic tasks. This thesis's general theme is to show that more human-like actuation enables exploring and failing directly on the real system while attempting fast and risky motions. In the first part of this thesis, we develop a robotic arm with four degrees of freedom and eight pneumatic artificial muscles (PAM).Such a system is capable of replicating desired behaviors as seen in human arm motions: 1) high power-to-weight ratios, 2) inherent robustness due to passive compliance and 3) high-speed catapult-like motions as possible with fast energy release. Rather than recreating human anatomy, this system is designed to simplify control than previously designed pneumatic muscle robots. One of the main insights is that a simple PID controller is sufficient to control this system for slow motions accurately. When exploring fast movements directly on the real system, the antagonistic actuation avoids damages to the system. In this manner, the PID controller's parameters and additional feedforward terms can be tuned automatically using Bayesian optimization without further safety considerations. Having such a system and following our goal to show the benefits of the combination of learning and muscular systems, the next part's content is to learn a dynamics model and use it for control. In particular, the goal here is to learn a model purely from data as analytical models of PAM-based robots are not sufficiently good. Nonlinearities, hysteresis effects, massive actuator delay, and unobservable dependencies like temperature make such actuators' modeling especially hard. We learn probabilistic forward dynamics models using Gaussian processes and, subsequently, employ them for control to address this issue. However, Gaussian processes dynamics models cannot be set-up for our musculoskeletal robot as for traditional motor-driven robots because of unclear state composition, etc. In this part of the thesis, we empirically study and discuss how to tune these approaches to complex musculoskeletal robots. For the control part, introduce Variance Regularized Control (VRC) that tracks a desired trajectory using the learned probabilistic model. VRC incorporates the GP's variance prediction as a regularization term to optimize for actions that minimize the tracking error while staying in the training data's vicinity. In the third part of this thesis, we utilized the PAM-based robot to return and smash table tennis balls that have been shot by a ball launcher. Rather than optimizing the desired trajectory and subsequently track it to hit the ball, we employ model-free Reinforcement Learning to learn this task from scratch. By using RL with our system, we can specify the table tennis task directly in the reward function. The RL agent also applies the actions directly on the low-level controls (equivalent to the air pressure space) while robot safety is assured due to the antagonistic actuation. In this manner, we allow the RL agent to be applied to the real system in the same way as in simulation. Additionally, we make use of the robustness of PAM-driven robots by letting the training run for 1.5 million time steps 14 hours. We introduce a semi sim and real training procedure in order to avoid training with real balls. With this solution, we return 75% of all incoming balls to the opponent's side of the table without using real balls during training. We also learn to smash the ball with an average ball speed of 12 m\\s (5 m\\s for the return task) after the hit while sacrificing accuracy (return rate of 29%). In summary, we show that learning approaches to control of muscular systems can lead to increased performance in dynamic tasks.", "source_url": "https:\/\/openalex.org\/W3120428840", "pdf_url": "https:\/\/tuprints.ulb.tu-darmstadt.de\/17210\/1\/phdthesis_buechler.pdf", "doi": "10.25534\/tuprints-00017210", "handle": "", "repository_name": "TUbilio (Technical University of Darmstadt)", "source_record_id": "W3120428840", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 541, "title": "TOWARDS TRUSTWORTHY NATURAL LANGUAGE PROCESSING", "title_ar": "", "title_en": "TOWARDS TRUSTWORTHY NATURAL LANGUAGE PROCESSING", "author": "HANNAH AUGHT SCHWARTZ BROWN", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Singapore", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "fairness", "categories": "nlp; responsible-ai; fairness", "keywords": "", "abstract": "", "source_url": "https:\/\/openalex.org\/W7163603768", "pdf_url": "", "doi": "", "handle": "", "repository_name": "National University of Singapore", "source_record_id": "W7163603768", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 547, "title": "Weak Supervision and Active Learning for Natural Language Processing", "title_ar": "", "title_en": "Weak Supervision and Active Learning for Natural Language Processing", "author": "MING LIU", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "nlp", "research_subdomain": "", "categories": "deep-learning; computer-vision; nlp", "keywords": "", "abstract": "Deep neural models in natural language processing rely on large amounts of labeled data. In the real world, annotation can be expensive and time consuming. In this project, the aim is to learn good deep neural models with a minimum amount of labeled data. We take multiple strategies to achieve this goal: we use structured information from the data, incorporate prior knowledge for the model and learn an active learning policy for annotation. Experiments on simulated and real word tasks show these strategies are useful and effective.", "source_url": "https:\/\/openalex.org\/W3124832285", "pdf_url": "https:\/\/figshare.com\/articles\/Weak_Supervision_and_Active_Learning_for_Natural_Language_Processing\/8396834", "doi": "10.26180\/5d1ab5f09f24b", "handle": "", "repository_name": "Figshare", "source_record_id": "W3124832285", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 557, "title": "Natural language processing as autoregressive generation", "title_ar": "", "title_en": "Natural language processing as autoregressive generation", "author": "Xiang Lin", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Nanyang Technological University", "university_source": "repository_host", "college": "", "department": "", "country": "Singapore", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; generative-ai; responsible-ai; predictive-analytics; llm; machine-translation; fairness", "keywords": "", "abstract": "The advances in deep learning have led to great achievements in many Natural Language Processing (NLP) tasks. With the nature of language, i.e., sequential data, most NLP tasks can be framed into the sequence learning framework, such as text generation. As one of the most important foundations for modern NLP techniques, autoregressive generation models have achieved dominant performance in a great deal of NLP tasks. Therefore, this thesis emphasizes improving the autoregressive generation model for different NLP tasks. While many tasks can naturally fit into the sequence learning framework, some of them, e.g., building discourse parsing tree, require sophisticated designs to fit into neural models. Therefore, this thesis firstly emphasizes a novel unified framework for discourse parsing, which builds a discourse tree in a top-down depth-first manner, and it frames the task as an autoregressive generation task with the goal of each step being the prediction of the node position given a piece of text. The proposed approach is proven effective with extensive empirical experiments. In addition, I extend the above framework by proposing a hierarchical decoder, which leverages the information from parents and siblings of the nodes that are currently processed. The proposed decoder utilizes the nature of the tree structure and further improves the experiment performance on both discourse parsing and dependency parsing tasks. On the other hand, the de facto strategies, i.e., cross entropy loss and teacher forcing, for training the autoregressive generation models have been shown problematic in certain aspects. For example, cross entropy loss, which is one of the widely leveraged training objective functions, often leads to text degeneration in text generation, and teacher forcing suffers from the exposure bias problem, where there exists a mismatch between the training and testing setup. For text degeneration, I introduce a class of diminishing attentions, which enforces the submodularity of the coverage calculated by cross attention in the sequence-to-sequence model. The proposed diminishing attentions achieve notable improvement on several neural text generation tasks, including text summarization, machine translation, and image paragraph generation. Further, I propose a novel training objective, ScaleGrad, to replace cross entropy, which significantly reduces the degeneration problem in different text generation tasks. In fact, ScaleGrad can be extended to problems beyond text degeneration. It provides wide flexibility to inject different inductive biases into the text generation model by directly modifying the gradient information in the output layer. Next, for the exposure bias problem, this thesis introduces a novel type of scheduled sampling based on training accuracy, which requires only minimal hyper-parameter tuning compared to existing scheduled sampling methods. Additionally, a novel imitation loss is proposed to further enforce the model’s generative behavior to match the teacher-forced behavior. Moreover, this thesis demonstrates that reducing exposure bias can improve the robustness of language models against repetition and toxic errors.", "source_url": "https:\/\/openalex.org\/W4382929605", "pdf_url": "", "doi": "10.32657\/10356\/168487", "handle": "", "repository_name": "DR-NTU (Nanyang Technological University)", "source_record_id": "W4382929605", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 558, "title": "Domain adaptation in Natural Language Processing", "title_ar": "", "title_en": "Domain adaptation in Natural Language Processing", "author": "Marina Sedinkina", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Ludwig-Maximilians-Universität München", "university_source": "repository_host", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "nlp", "research_subdomain": "sentiment-analysis", "categories": "machine-learning; nlp; transfer-learning; predictive-analytics; sentiment-analysis", "keywords": "", "abstract": "Domain adaptation has received much attention in the past decade. It has been shown that domain knowledge is paramount for building successful Natural Language Processing (NLP) applications. To investigate the domain adaptation problem, we conduct several experiments from different perspectives. First, we automatically adapt sentiment dictionaries for predicting the financial outcomes “excess return” and “volatility”. In these experiments, we compare manual adaptation of the domain-general dictionary with automatic adaptation, and manual adaptation with a combination consisting of first manual, then automatic adaptation. We demonstrate that automatic adaptation performs better than manual adaptation, namely the automatically adapted sentiment dictionary outperforms the previous state of the art in predicting excess return and volatility. Furthermore, we perform qualitative and quantitative analyses finding that annotation based on an expert’s a priori belief about a word’s meaning is error-prone – the meaning of a word can only be recognized in the context that it appears in. Second, we develop the temporal transfer learning approach to account for the language change in social media. The language of social media is changing rapidly – new words appear in the vocabulary, and new trends are constantly emerging. Temporal transfer-learning allows us to model these temporal dynamics in the document collection. We show that this method significantly improves the prediction of movie sales from discussions on social media forums. In particular, we illustrate the success of parameter transfer, the importance of textual information for financial prediction, and show that temporal transfer learning can capture temporal trends in the data by focusing on those features that are relevant in a particular time step, i.e., we obtain more robust models preventing overfitting. Third, we compare the performance of various domain adaptation models in low-resource settings, i.e., when there is a lack of large amounts of high-quality training data. This is an important issue in computational linguistics since the success of NLP applications primarily depends on the availability of training data. In real-world scenarios, the data is often too restricted and specialized. In our experiments, we evaluate different domain adaptation methods under these assumptions and find the most appropriate techniques for such a low-data problem. Furthermore, we discuss the conditions under which one approach substantially outperforms the other. Finally, we summarize our work on domain adaptation in NLP and discuss possible future work topics.", "source_url": "https:\/\/openalex.org\/W3164246911", "pdf_url": "", "doi": "10.5282\/edoc.27831", "handle": "", "repository_name": "Elektronische Hochschulschriften der LMU München (Ludwig-Maximilians-Universität München)", "source_record_id": "W3164246911", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 564, "title": "Self- and semi-supervised machine learning techniques for discovering novel ocular and neuroimaging biomarkers of Alzheimer's disease", "title_ar": "", "title_en": "Self- and semi-supervised machine learning techniques for discovering novel ocular and neuroimaging biomarkers of Alzheimer's disease", "author": "Charlene Z. L. Ong", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Nanyang Technological University", "university_source": "repository_host", "college": "", "department": "", "country": "Singapore", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; generative-ai; medical-ai; responsible-ai; self-semi-supervised; predictive-analytics; cnn; gnn; multimodal; image-segmentation; gan; medical-imaging; xai", "keywords": "", "abstract": "There is increasing prevalence of Alzheimer’s disease worldwide. To manage the progression of Alzheimer’s disease, biomarkers can be utilized to identify patients with the disease or predict subsequent progression. While neuroimaging biomarkers have been long established to visualize structural changes associated with Alzheimer’s disease, there are other promising biomarkers such as retinal imaging biomarkers which offer the possibility of detecting early pathophysiological changes and are non-invasive and widely available. Nevertheless, the link between retinal imaging biomarkers to Alzheimer’s disease remains unclear and there is much difficulty in procuring datasets of patients for investigation, in particular longitudinal datasets. At the same time, advances in deep learning also offer much promise for furthering our understanding of the link between ophthalmic biomarkers and dementia. Self-supervised learning techniques offer potential for learning new biomarkers which might not have been defined previously. However, there is often a lack of sufficient ground truth annotations, hence hampering the application of deep learning models. In light of this, the work in this thesis aims to address these issues by exploring the use of self or semi-supervised learning approaches and leveraging on both neuroimaging and ophthalmic imaging. By utilizing unlabeled datasets, self or semi-supervised learning could potentially improve the model performance on a range of tasks, such as segmentation, disease classification, and prognosis prediction. More importantly, self-supervised learning techniques could uncover novel biomarkers in a data-driven manner. Some of these methods include generative adversarial networks, graph neural networks, and masked autoencoders, which are explored in this study for biomarker discovery. Firstly, the challenge of lack of sufficient ground truth annotation to develop models for retinal layer segmentation is investigated. Retinal layer segmentation is critical for quantifying the retinal layer thickness, which has been shown to be a useful diagnostic biomarker for age-related macular degeneration and a potential biomarker for Alzheimer’s disease. While there have been existing approaches developed which reduce annotation need, we propose to develop a deep learning approach using generative adversarial learning that utilizes a very limited number of labels for segmentation. We demonstrated that our approach showed comparable or superior performance to other state-of-the-art approaches, particularly with limited labels. In addition, we also evaluated our approach under various noisy conditions and showed it was robust in most noisy conditions. Secondly, current dementia diagnosis is based primarily on clinical assessment. Neuroimaging in the form of magnetic resonance imaging is used to provide additional granularity. It will hence be practical to explore neuroimaging biomarkers as a value add to clinical assessment, especially in a pre-dementia cohort. Our proposed approach uses a graph convolutional neural network to combine both clinical and non-clinical features such as neuroimaging for the classification of cognitive syndromes. It integrates a Cross Attention Module to highlight the salient features in the non-clinical features, attaining the highest accuracy and Matthew’s correlation coefficient compared to existing state-of-the-art methods. In addition, by investigating the feature interpretability via Integrated Gradients, several nonclinical features, such as cerebrospinal fluid volume, right hippocampal fissure volume, right frontal pole thickness, and left pallidum volume were identified as key features for the classification. Thirdly, we explore the utility of ophthalmic biomarkers as prognostic biomarkers for cognitive decline in Alzheimer’s disease patients. In parallel, we explore the possibility of combining both ophthalmic and neuroimaging features to form multimodal features. The application of these multimodal features is investigated in a clinical trial dataset consisting of participants with mild-to-moderate Alzheimer’s disease. To learn key feature representations associated with the ophthalmic images, we utilize self-supervised learning via masked autoencoders to extract feature representations. For neuroimaging, due to the availability of established neuroimaging software, we utilize standard preprocessing pipelines to obtain these neuroimaging features. We investigated the combination of these features using a graph convolutional network model for the prediction of cognitive decline in Alzheimer’s disease and showed initial promising results with the addition of optical coherence tomography data.", "source_url": "https:\/\/openalex.org\/W7127051058", "pdf_url": "", "doi": "10.32657\/10356\/205534", "handle": "", "repository_name": "DR-NTU (Nanyang Technological University)", "source_record_id": "W7127051058", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 565, "title": "Semi-Self-Supervised Learning for Deep Models: Reducing Annotation Costs in Dense-Pattern Agricultural Image Analysis", "title_ar": "", "title_en": "Semi-Self-Supervised Learning for Deep Models: Reducing Annotation Costs in Dense-Pattern Agricultural Image Analysis", "author": "Najafian, Keyhan", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Saskatchewan", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; deep-learning; computer-vision; generative-ai; ai-applications; transfer-learning; self-semi-supervised; object-detection; image-segmentation; video-analysis; ai-agriculture", "keywords": "", "abstract": "Most of the success of deep learning in computer vision depends on supervised learning (SL) and large-scale, manually annotated datasets. Creating such datasets is a major bottleneck due to the substantial cost and labor required, particularly for pixel-level tasks on images with dense and irregular object distributions, as commonly encountered in precision agriculture. While existing semi- and self-supervised methods attempt to reduce this dependency by requiring minimal to no annotation, they still require massive, difficult-to-acquire unlabeled datasets. Self-supervised learning typically involves pretraining on extensive unlabeled data to learn generalizable representations, followed by fine-tuning on task-specific labeled samples, whereas semi-supervised learning combines a small set of labeled data with larger unlabeled collections. To mitigate these challenges, this dissertation introduces a semi-self-supervised learning paradigm that integrates data synthesis, domain adaptation, and spatiotemporal modeling to develop high-performing models across various tasks. Specifically, our paradigm leverages high-frame-rate unannotated video, supplemented with minimal human-annotated static images and video frames, to create large-scale annotated datasets more efficiently than conventional data collection. The produced data includes synthetic training samples, which are rendered through data simulation (synthetic data), and automatically labeled real samples (generated data), where all labels originate from the initial minimal human annotations, thus eliminating the need for further manual labeling and directly addressing the data scarcity problem. We demonstrate the efficacy of our proposed semi-self-supervised paradigm through four key studies on progressively more challenging computer vision tasks within the demanding context of precision agriculture. The framework is first applied to object detection, where a computational synthesis method creates large-scale bounding-box annotated datasets from minimal manually annotated video frames, achieving robust performance of over 80%. Extending this to the pixel-level, the method generates high-quality semantic masks for scenes characterized by dense and overlapping objects, reaching a 91% Dice score with merely 19 annotated video frames and static images. For the more granular task of instance segmentation, our approach enables scalable instance-level object analysis, obtaining over 90% accuracy. Finally, by integrating optical flow and spatiotemporal modeling, our method effectively segments objects in complex video sequences with minimal supervision, outperforming conventional approaches. Together, these studies validate the framework’s ability to deliver high-performing models across diverse and complex tasks while drastically reducing manual labeling effort. These studies establish that our proposed semi-self-supervised paradigm offers a transformative approach for developing DL models. By effectively leveraging video data and supervision derived from minimal human annotations, coupled with domain adaptation, this methodology demonstrates significant scalability and generalizability. It successfully enables instance-level and pixel-level reasoning even with limited annotation, confirming its potential to overcome annotation bottlenecks and advance DL applications in complex, data-constrained real-world scenarios, particularly within the field of precision agriculture.", "source_url": "https:\/\/openalex.org\/W7113662752", "pdf_url": "", "doi": "", "handle": "", "repository_name": "HARVEST (University of Saskatchewan)", "source_record_id": "W7113662752", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 568, "title": "Improving Human Behavior Understanding with Semi-supervised Learning", "title_ar": "", "title_en": "Improving Human Behavior Understanding with Semi-supervised Learning", "author": "Bruno Manuel Degardin", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Portugal", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; nlp; generative-ai; responsible-ai; self-semi-supervised; cnn; transformers; multimodal; video-analysis; gan; diffusion; fairness", "keywords": "", "abstract": "Recent advancements in Computer Vision (CV) and Machine Learning (ML) have fundamentally revolutionized the field of human behaviour understanding, enabling intelligent systems to move beyond simple motion detection to the interpretation of complex, semantic actions from vast streams of visual data. This paradigm shift has been driven by the exponential growth of Deep Learning algorithms, particularly the evolution from 3D Convolutional Neural Networks (3D CNNs) to modern Vision Transformers (ViTs) , which have achieved remarkable success in modeling long-range spatiotemporal dependencies in RGB videos and multimodal signals. However, despite their impressive performance on controlled benchmarks, the deployment of these pixel-based models in real-world scenarios remains heavily constrained. Their reliance on high-dimensional visual data often incurs prohibitive computational costs that hinder real-time application on edge devices. Furthermore, these methods exhibit significant sensitivity to background clutter and viewpoint variations , and the processing of raw video raises critical privacy concerns by capturing unnecessary appearance details such as identity and clothing. Maximizing the potential of this modality presents a significant set of challenges that hinder widespread deployment. The heavy reliance on massive, high-quality annotated datasets creates a prohibitive bottleneck for supervised learning, while existing generative models used for data augmentation often lack physical plausibility or controllability. Furthermore, self-supervised methods borrowed from video domains frequently fail to capture the fine-grained dynamics of skeletal motion, and current evaluation metrics for synthetic data struggle to distinguish between realistic kinematics and artifacts. Addressing these limitations requires a fundamental shift toward more label-efficient, physically aware, and generative approaches to behaviour analysis. In this thesis, we proposed a comprehensive suite of novel solutions to mitigate these problems by exploiting semi-supervised and self-supervised learning paradigms, as well as advanced generative modeling. First, we addressed the limitations of temporal reasoning in Graph Convolutional Networks (GCNs) by proposing REGINA, a reasoning- augmented architecture. This method injects explicit temporal self-similarity cues into the network through a learnable temporal adjacency matrix, allowing the model to adaptively optimize its receptive fields to capture long-range dependencies that standard convolutions often miss. Addressing data scarcity, we proposed Kinetic-GAN, a generative adversarial network that scales action conditioning by an order of magnitude over prior works. By generating actions directly from a disentangled latent space using a mapping network and stochastic noise injection, it synthesizes diverse, high-fidelity global body movements across more than 100 action classes. We further advanced representation learning without labels through ATOM, a self-supervised strategy that leverages atomic motion representation learning. By decomposing actions into short, fine-grained primitives and training the backbone to recover their temporal order, ATOM learns robust features from the intrinsic structure of motion, transferring effectively to supervised tasks without the computational burden of heavy reconstruction heads. To bridge the gap between generation and recognition, we developed an intrinsic, kinematics-aware quality assessment framework. Recognizing that conventional distribution metrics fail to capture physical validity, this framework utilizes per-sample kinematic indicators within a weaklysupervised discriminator to iteratively filter synthetic samples, ensuring only physically plausible data is used to boost downstream performance. Finally, pushing the boundaries of controllable synthesis, we proposed GLDM, a Spatiotemporal Graph Latent Diffusion Model. GLDM projects skeleton sequences into a compact, graph-aligned latent space and performs diffusion with graph-aware updates. Crucially, by integrating cross-attention mechanisms, GLDM enables precise multi-modal conditioning, allowing generation to be guided not just by class labels but also by natural language prompts, achieving generalization beyond training data, unconstrained temporal flexibility and solving the mode collapse issues inherent to adversarial approaches. Results obtained by the proposed methods show that it is possible to improve HBU systems regarding label efficiency, generative controllability, and physical plausibility, while still obtaining state-of-the-art results. All proposed methods were evaluated on several large-scale benchmarks, demonstrating that robust behaviour understanding can be achieved with significantly reduced reliance on manual annotation.", "source_url": "https:\/\/openalex.org\/W7225277943", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT)", "source_record_id": "W7225277943", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 569, "title": "Efficient Rare Disease Detection with Self-Supervised and Transfer Learning", "title_ar": "", "title_en": "Efficient Rare Disease Detection with Self-Supervised and Transfer Learning", "author": "Miftahul Jannat", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Eastern Finland", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; transfer-learning; self-semi-supervised; predictive-analytics; cnn; medical-imaging", "keywords": "", "abstract": "Accurate diagnosis of rare diseases by analyzing medical images is often faced with the problem of limited availability of annotated datasets, as the manual annotation process, being time-consuming, requires expert knowledge in addition to a significant time investment.The research explores the use of Self-Supervised Learning (SSL) combined with Transfer Learning as a method to reduce the dependency on labeled data without compromising diagnostic performance.The approach designed involves pretraining convolutional neural networks using large unlabeled datasets through self-supervised learning (SSL) methods.For this purpose, the CheXpert dataset of over 200000 X-ray images is employed.To enable the model to imbibe generalizable features from the unlabeled pictures, several SSL pretext tasks are applied, namely rotation prediction, jigsaw puzzles, as well as contrastive learning (SimCLR).Following the pretraining process, the models are fine-tuned using the ISIC 2018 dataset, which contains annotated dermoscopic images for the task of skin lesion classification.Feature extraction and classification performance are evaluated using different architectures, such as ResNet-50.Finetuning is performed using different percentages of labeled data (10%, 25%, 50%, and 100%) to evaluate the efficiency of annotations in semi-supervised learning (SSL).The performance and models have been assessed with respect to Accuracy, Precision, Recall, and F1 Score.From the results, it appears that SSL-pretrained models using SimCLR outperformed fully supervised models even when they were trained on significantly fewer labeled samples.It is astonishing that models which were trained with only 50% of the labeled data actually achieved over 75% accuracy in comparison to those trained on the complete dataset.This work demonstrates that the semi-supervised learning (SSL) approach made possible through transfer learning is a highly effective and efficient in terms of labeling resources for detecting rare diseases in medical images.", "source_url": "https:\/\/openalex.org\/W7215667449", "pdf_url": "https:\/\/erepo.uef.fi\/bitstreams\/9b495e9a-88c8-4914-8f52-550745585eb9\/download", "doi": "", "handle": "", "repository_name": "UEF eRepo (University of Eastern Finland)", "source_record_id": "W7215667449", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 571, "title": "Semi-supervised learning using pseudo-labels: A case study in Northern Sámi ASR", "title_ar": "", "title_en": "Semi-supervised learning using pseudo-labels: A case study in Northern Sámi ASR", "author": "Pal, Priyanshi", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Aalto University", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "machine-learning; nlp; ai-applications; transfer-learning; self-semi-supervised; speech-recognition; ai-education", "keywords": "", "abstract": "While Self-Supervised Learning has advanced Automatic Speech Recognition (ASR) for low-resource languages, limited labeled data remains a bottleneck, often preventing reasonable downstream performance. In line with this requirement, Semi-Supervised Learning approaches have gained prominence as providing a means to data-efficient methods that can be helpful, even with limited supervised data. This work investigates the usefulness of semi-supervised learning, specifically Pseudo-Labeling with self-supervised speech models for Northern Sámi, an under-resourced agglutinative language. Within a teacher–student framework, pseudo-labels are incorporated using two strategies: (a) filtering them through WER-based agreement between the models, and (b) using all pseudo-labels generated by a teacher model. Results show that pseudo-labels from the larger teacher foundation model can improve the downstream performance of a smaller one without additional labeled data, yielding 33.30–3.84% relative character error rate reduction (CERR) across three out-of-distribution test sets. The smaller student model is evaluated in a supervised setting (20 hours of labeled data) and semi-supervised settings with 20, 26 and 75 hours of pseudo-labeled data. Notably, the student model can benefit as much, or even more, from pseudo-labels as from labeled data. Further gains can be achieved when both annotated and pseudo-labels are combined in a two-stage fine-tuning process, rather than mixing all data in a single training phase.", "source_url": "https:\/\/openalex.org\/W7115821482", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Aaltodoc (Aalto University)", "source_record_id": "W7115821482", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 572, "title": "Transformers and contrastive semi-supervised learning for medical image segmentation", "title_ar": "", "title_en": "Transformers and contrastive semi-supervised learning for medical image segmentation", "author": "Qianying Liu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "The University of Glasgow", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; responsible-ai; ai-applications; self-semi-supervised; cnn; transformers; image-segmentation; medical-imaging; fairness; ai-education", "keywords": "", "abstract": "Medical Image Semantic Segmentation (MISS), the process of assigning a semantic label to each pixel in an image, is a foundational task in computational medicine, critical for quantitative diagnostics and treatment planning. However, developing robust MISS models faces two intertwined challenges. First, there is an architectural dilemma: Convolutional Neural Networks (CNNs), like U-Net, excel at learning local features but are limited by their receptive fields, failing to capture global context essential for segmenting organs with large deformations. Conversely, Vision Transformers (ViTs) effectively model long-range dependencies but lack the inductive biases of CNNs, leading to poor generalization on the small datasets typical in medicine without extensive pre-training. Second, the prohibitive cost and expertise required for creating pixel-level annotations create a severe data scarcity bottleneck. While Semi-Supervised Learning (SSL) aims to mitigate this by leveraging unlabeled data, existing methods often fail to learn high-level semantic relations and are susceptible to confirmation bias from noisy pseudo-labels, class imbalance, and suboptimal contrastive sample selection. This thesis presents a comprehensive investigation to systematically address these challenges, delivering a cohesive suite of novel deep learning frameworks. The contributions are four-fold: First, to resolve the architectural trade-off, this work introduces CS-Unet, a pure Transformer network built upon a U-Net-like architecture. Its core innovation is the Convolutional Swin Transformer (CST) block, which integrates convolutions directly within the Multi-Head Self-Attention and Feed-Forward Network modules. This design imbues the Transformer with inherent localized spatial context and strong inductive biases, enabling it to efficiently learn both local and global features. Without pre-training, CS-Unet outperforms existing Transformer and CNN-based models on multi-organ and cardiac datasets, achieving state-of-the-art performance with fewer parameters. Second, to address data scarcity, a novel Multi-Scale Cross Supervised Contrastive Learning (MCSC) framework for SSL is developed. MCSC jointly trains CNN and Transformer models, using a cross-teaching paradigm where each network provides pseudo-labels for the other. Crucially, it moves beyond simple output consistency by applying a contrastive loss to feature maps at multiple scales, enforcing hierarchical semantic consistency. To handle the class imbalance endemic to medical imaging, a class-prevalence-aware loss is used to ensure features for infrequent classes are learned robustly. Third, to fortify SSL against noisy pseudo-labels, a certainty-guided contrastive learning strategy is proposed. This approach mitigates the impact of inaccurate pseudo-labels by using a certainty metric to guide the selection of samples for contrastive learning. The framework’s computational efficiency is enhanced through novel sampling strategies that select a few representative samples for contrasting, and a negative memory bank is used to increase sample diversity and eliminate dependence on batch size. Fourth, this thesis introduces a new paradigm for SSL by leveraging external anatomical priors through the Contrastive Cross-Teaching with Registration (CCT-R) framework. CCT-R is the first method to integrate spatial registration transforms into the learning process. It features two novel modules: a Registration Supervision Loss (RSL), which uses transforms between labeled and unlabeled volumes to generate an additional, highly reliable source of pseudo-labels, and Registration-Enhanced Positive Sampling (REPS), which uses registration to identify anatomically-corresponding positive pairs across volumes for contrastive learning. Overall, these contributions provide a powerful toolkit that significantly alleviates the annotation bottleneck in medical AI. The proposed methods demonstrate state-of-the-art performance on challenging segmentation benchmarks, delivering a pathway to develop accurate, data-efficient models for real-world clinical applications and opening new avenues for research into fusing geometric priors with semantic segmentation.", "source_url": "https:\/\/openalex.org\/W7127166515", "pdf_url": "", "doi": "10.5525\/gla.thesis.85734", "handle": "", "repository_name": "Enlighten: Theses (The University of Glasgow)", "source_record_id": "W7127166515", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 575, "title": "Self-Supervised Learning for Geometry", "title_ar": "", "title_en": "Self-Supervised Learning for Geometry", "author": "Huangying Zhan", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Adelaide", "university_source": "repository_host", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "video-analysis", "categories": "machine-learning; deep-learning; computer-vision; robotics; ai-applications; self-semi-supervised; predictive-analytics; video-analysis; ai-smart-cities", "keywords": "", "abstract": "This thesis focuses on two fundamental problems in robotic vision, scene geometry understanding and camera tracking. While both tasks have been the subject of research in robotic vision, numerous geometric solutions have been proposed in the past decades. In this thesis, we cast the geometric problems as machine learning problems, specifically, deep learning problems. Differ from conventional supervised learning methods that using expensive annotations as the supervisory signal, we advocate for the use of geometry as a supervisory signal to improve the perceptual capabilities in robots, namely Geometry Self-supervision. With the geometry self-supervision, we allow robots to learn and infer the 3D structure of the scene and ego-motion by watching videos, instead of expensive ground-truth annotation in traditional supervised learning problems. Followed by showing the use of geometry for deep learning, we show the possibilities of integrating self-supervised models with traditional geometry-based methods as a hybrid solution for solving the mapping and tracking problem. We focus on an end-to-end mapping problem from stereo data in the first part of this thesis, namely Deep Stereo Matching. Stereo matching is one of the oldest problems in computer vision. Classical approaches to stereo matching typically rely on handcrafted features and a multiple-step solution. Recent deep learning methods utilize deep neural networks to achieve end-to-end trained approaches while significantly outperforming classic methods. We propose a novel data acquisition pipeline using an untethered device (Microsoft HoloLens) with a Time-of-Flight (ToF) depth camera and stereo cameras to collect real-world data. A novel semi-supervised method is proposed to train networks with ground-truth supervision and self-supervision. The large scale real-world stereo dataset with semi-dense annotation and dense self-supervision allow our deep stereo matching network to generalize better when compared to prior arts. Mapping and tracking using a single camera (Monocular) is a harder problem when compared to that using a stereo camera due to varies well-known challenges. In the second part of this thesis, We decouple the problem into single view depth estimation (mapping) and two view visual odometry (tracking) and propose a self-supervised framework, namely SelfTAM, which jointly learns the depth estimator and the odometry estimator. The self-supervised problem is usually formulated as an energy minimization problem consist of an energy of data consistency in multi-view (e.g. photometric) and an energy of prior regularization (e.g. depth smoothness prior). We strengthen the supervision signal with a deep feature consistency energy term and a surface normal regularization term. Though our method trains models with stereo sequence such that a real-world scaling factor is naturally incorporated, only monocular data is required in the inference stage. In the last part of this thesis, we revisit the basics of visual odometry and explore the best practice to integrate deep learning models with geometry-based visual odometry methods. A robust visual odometry system, DF-VO, is proposed. We use deep networks to establish 2D-2D\/3D-2D correspondences and pick the best correspondences from the dense predictions. Feeding the high-quality correspondences into traditional VO methods, e.g. Epipolar Geometry and Prospective-n-Points, we can solve visual odometry problem within a more robust framework. With the proposed self-supervised training, we can even allow the models to perform online adaptation in the run-time and take a step toward a lifelong learning visual odometry system.", "source_url": "https:\/\/openalex.org\/W3132493307", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Adelaide Research & Scholarship (AR&S) (University of Adelaide)", "source_record_id": "W3132493307", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 576, "title": "Optimization Methods for Semi-Supervised Learning", "title_ar": "", "title_en": "Optimization Methods for Semi-Supervised Learning", "author": "Edward Cheung", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Waterloo", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "medical-ai", "research_subdomain": "self-semi-supervised", "categories": "machine-learning; medical-ai; self-semi-supervised; predictive-analytics; optimization", "keywords": "", "abstract": "The goal of this thesis is to provide efficient optimization algorithms for some semi-supervised learning (SSL) tasks in machine learning. For many machine learning tasks, training a classifier requires a large amount of labeled data; however, providing labels typically requires costly manual annotation. Fortunately, there is typically an abundance of unlabeled data that can be easily collected for many domains. In this thesis, we focus on problems where an underlying structure allows us to leverage the large amounts of unlabeled data, while only requiring small amounts of labeled data. In particular, we consider low-rank matrix completion problems with applications to recommender systems, and semi-supervised support vector machines (S3VM) to solve binary classification problems, such as digit recognition or disease classification. For the first class of problems, we study convex approximations to the low-rank matrix completion problem. Instead of restricting the solution space to low-rank matrices, we use the trace norm as a convex surrogate. Unfortunately, many trace norm minimization algorithms scale very poorly in practice since they require a full singular value decomposition (SVD) at each iteration. Recently, there has been renewed interest in the trace norm constrained problem utilizing the Frank-Wolfe algorithm, which only requires calculating the leading singular vector pair, providing an order of magnitude improvement on the iteration complexity. However, the Frank-Wolfe algorithm empirically has very slow convergence and in practice yields high-rank solutions, which greatly increases computational costs. To address this issue, we investigate a rank-drop step for Frank-Wolfe, which solves a subproblem specifically designed to decrease the rank of the iterate, ensuring that the Frank-Wolfe algorithm converges along a low-rank path. We show that this rank-drop subproblem can be decomposed into two cases, where each subproblem can be solved efficiently and we guarantee that the iterates remain feasible, preserving the projection-free property of Frank-Wolfe. Next we show that these ideas can be used to provide scalable algorithms for simultaneously sparse and low-rank matrix completion problems. We extend the Frank-Wolfe analysis to accommodate nonsmooth objectives, which can be used to solve the simultaneously sparse and low-rank problem. We replace the traditional linear approximation used in Frank-Wolfe by a uniform affine approximation to better address poor local approximations given by the first-order Taylor approximation. We show that this naturally leads to a sequence of smooth functions that uniformly converges to the original nonsmooth objective, allowing for a careful balance between approximation quality and convergence that is closely related to the step sizes of the Frank-Wolfe algorithm. We apply this algorithm to solve sparse covariance estimation problems, graph link prediction, and robust matrix completion problems. Finally, we propose a variant of self-training for the semi-supervised binary classification problem by leveraging ideas from S3VM. To address common issues associated with self-training, such as error propagation and label imbalances, we proposed an adaptive scheme using the functional margin of S3VM to construct a confidence measure. The confidence score is used to create rules to adapt the optimization problems to incorporate label uncertainty and class imbalances. Moreover, we show that the incremental training approach leverages warm-starts very well, leading to much faster training than standard S3VM methods alone, with much stronger empirical performance on imbalanced datasets.", "source_url": "https:\/\/openalex.org\/W2803599927", "pdf_url": "", "doi": "", "handle": "", "repository_name": "UWSpace (University of Waterloo)", "source_record_id": "W2803599927", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 579, "title": "Semi-Supervised Hybrid Windowing Ensembles for Learning from Evolving Streams", "title_ar": "", "title_en": "Semi-Supervised Hybrid Windowing Ensembles for Learning from Evolving Streams", "author": "Sean Louis Alan Floyd", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Ottawa", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; computer-vision; self-semi-supervised; predictive-analytics; object-detection", "keywords": "", "abstract": "In this thesis, learning refers to the intelligent computational extraction of knowledge from data. Supervised learning tasks require data to be annotated with labels, whereas for unsupervised learning, data is not labelled. Semi-supervised learning deals with data sets that are partially labelled. A major issue with supervised and semi-supervised learning of data streams is late-arriving or missing class labels. Assuming that correctly labelled data will always be available and timely is often unfeasible, and, as such, supervised methods are not directly applicable in the real world. Therefore, real-world problems usually require the use of semi-supervised or unsupervised learning techniques. For instance, when considering a spam detection task, it is not reasonable to assume that all spam will be identified (correctly labelled) prior to learning. Additionally, in semi-supervised learning, \"the instances having the highest [predictive] confidence are not necessarily the most useful ones\" [41]. We investigate how self-training performs without its selective heuristic in a streaming setting. This leads us to our contributions. We extend an existing concept drift detector to operate without any labelled data, by using a sliding window of our ensemble's prediction confidence, instead of a boolean indicating whether the ensemble's predictions are correct. We also extend selective self-training, a semi-supervised learning method, by using all predictions, and not only those with high predictive confidence. Finally, we introduce a novel windowing type for ensembles, as sliding windows are very time consuming and regular tumbling windows are not a suitable replacement. Our windowing technique can be considered a hybrid of the two: we train each sub-classifier in the ensemble with tumbling windows, but delay training in such a way that only one sub-classifier can update its model per iteration. We found, through statistical significance tests, that our framework is (roughly 160 times) faster than current state of the art techniques, and achieves comparable predictive accuracy. That being said, more research is needed to further reduce the quantity of labelled data used for training, while also increasing its predictive accuracy.", "source_url": "https:\/\/openalex.org\/W2954664475", "pdf_url": "", "doi": "10.20381\/ruor-23520", "handle": "", "repository_name": "uO Research (University of Ottawa)", "source_record_id": "W2954664475", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 580, "title": "Semi-supervised dictionary learning and Semi-supervised deep neural network", "title_ar": "", "title_en": "", "author": "Khanh-Hung Tran", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2021, "language": "fr", "research_domain": "computer-vision", "research_subdomain": "self-semi-supervised", "categories": "machine-learning; deep-learning; computer-vision; robotics; ai-applications; self-semi-supervised; ai-cybersecurity", "keywords": "", "abstract": "Apprentissage semi-supervisé de dictionnaire et de réseaux de neurones profonds Depuis les années 2010, l’apprentissage automatique (ML) est l’un des sujets qui retient beaucoup l'attention des chercheurs scientifiques. De nombreux modèles de ML ont démontré leur capacité produire d’excellent résultats dans des divers domaines comme Vision par ordinateur, Traitement automatique des langues, Robotique… Toutefois, la plupart de ces modèles emploient l’apprentissage supervisé, qui requiert d’un massive annotation. Par conséquent, l’objectif de cette thèse est d’étudier et de proposer des approches semi-supervisées qui ont plusieurs avantages par rapport à l’apprentissage supervisé. Au lieu d’appliquer directement un classificateur semi-supervisé sur la représentation originale des données, nous utilisons plutôt des types de modèle qui intègrent une phase de l’apprentissage de représentation avant de la phase de classification, pour mieux s'adapter à la non linéarité des données. Dans le premier temps, nous revisitons des outils qui permettent de construire notre modèles semi-supervisés. Tout d’abord, nous présentons deux types de modèle qui possèdent l’apprentissage de représentation dans leur architecture : l’apprentissage de dictionnaire et le réseau de neurones, ainsi que les méthodes d’optimisation pour chaque type de model, en plus, dans le cas de réseau de neurones, nous précisons le problème avec les exemples contradictoires. Ensuite, nous présentons les techniques qui accompagnent souvent avec l’apprentissage semi-supervisé comme l’apprentissage de variétés et le pseudo-étiquetage. Dans le deuxième temps, nous travaillons sur l’apprentissage de dictionnaire. Nous synthétisons en général trois étapes pour construire un modèle semi-supervisée à partir d’un modèle supervisé. Ensuite, nous proposons notre modèle semi-supervisée pour traiter le problème de classification typiquement dans le cas d’un faible nombre d’échantillons d’entrainement (y compris tous labellisés et non labellisés échantillons). D'une part, nous appliquons la préservation de la structure de données de l’espace original à l’espace de code parcimonieux (l’apprentissage de variétés), ce qui est considéré comme la régularisation pour les codes parcimonieux. D'autre part, nous intégrons un classificateur semi-supervisé dans l’espace de code parcimonieux. En outre, nous effectuons le codage parcimonieux pour les échantillons de test en prenant en compte aussi la préservation de la structure de données. Cette méthode apporte une amélioration sur le taux de précision par rapport à des méthodes existantes. Dans le troisième temps, nous travaillons sur le réseau de neurones. Nous proposons une approche qui s’appelle \"manifold attack\" qui permets de renforcer l’apprentissage de variétés. Cette approche est inspirée par l’apprentissage antagoniste : trouver des points virtuels qui perturbent la fonction de coût sur l’apprentissage de variétés (en la maximisant) en fixant les paramètres du modèle; ensuite, les paramètres du modèle sont mis à jour, en minimisant cette fonction de coût et en fixant les points virtuels. Nous fournissons aussi des critères pour limiter l’espace auquel les points virtuels appartiennent et la méthode pour les initialiser. Cette approche apporte non seulement une amélioration sur le taux de précision mais aussi une grande robustesse contre les exemples contradictoires. Enfin, nous analysons des similarités et des différences, ainsi que des avantages et inconvénients entre l’apprentissage de dictionnaire et le réseau de neurones. Nous proposons quelques perspectives sur ces deux types de modèle. Dans le cas de l’apprentissage de dictionnaire semi-supervisé, nous proposons quelques techniques en inspirant par le réseau de neurones. Quant au réseau de neurones, nous proposons d’intégrer \"manifold attack\" sur les modèles génératifs.", "source_url": "https:\/\/openalex.org\/W7149497300", "pdf_url": "https:\/\/theses.hal.science\/tel-03206023v1\/document", "doi": "10.70675\/42000ac0z65ebz4a00z9dd3z6af0c02d8611", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7149497300", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 583, "title": "Semi-supervised learning methods for unsupervised domain adaptation in medical imaging segmentation", "title_ar": "", "title_en": "Semi-supervised learning methods for unsupervised domain adaptation in medical imaging segmentation", "author": "Pedro L. Ballester", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; computer-vision; medical-ai; transfer-learning; self-semi-supervised; predictive-analytics; image-segmentation; medical-imaging", "keywords": "", "abstract": "Machine learning applications make several assumptions regarding the scenario where they are employed. One common assumption is that data distribution in the test environment follows the same distribution of the training set. This assumption is systematically broken in most real-world scenarios; the difference between these distributions is commonly known as domain shift. Unsupervised domain adaptation aims at suppressing this problem by leveraging knowledge with unlabeled data from the test environment. One of the most sensitive fields for domain shift is medical imaging. Due to the heterogeneity in data distributions from scanners, models tend to vary in predictive performance when dealing with images from scanners with no examples in the training set. We propose two contributions in this work. First, we introduce the use of self-ensembling domain adaptation in the field of medical imaging segmentation in a spinal cord grey matter segmentation task. Next, based on the success of self-ensembling, we adapt two other recent work from the semi-supervised learning literature to the same task, namely, unsupervised data augmentation and MixMatch. We conduct ablation studies and other experiments in order to understand the behavior of each method and compare their best results. The results show improvements over training models in a supervised learning fashion and demonstrate that recent semi-supervised learning methods are promising for domain adaptation in medical imaging segmentation.", "source_url": "https:\/\/openalex.org\/W2992030155", "pdf_url": "", "doi": "", "handle": "", "repository_name": "LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas)", "source_record_id": "W2992030155", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 584, "title": "Semi-Supervised Learning with Unlabeled data: from Centralized to Distributed Systems", "title_ar": "", "title_en": "Semi-Supervised Learning with Unlabeled data: from Centralized to Distributed Systems", "author": "Nan Yang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "The University of Sydney", "university_source": "repository_host", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "computer-vision", "research_subdomain": "image-classification", "categories": "machine-learning; deep-learning; computer-vision; responsible-ai; self-semi-supervised; cnn; image-classification; federated-learning; fairness", "keywords": "", "abstract": "The rapid increase in data generated by edge devices and IoT technologies demands efficient management solutions, especially in terms of cost and infrastructure. Key challenges include expensive data labeling, requiring significant human resources, and ensuring data privacy and security, with the risk of information leakage during transmission. The limited availability of labeled data versus the exponential growth of new data presents challenges for maintaining accuracy and efficiency in data-driven models. High costs of data annotation, especially involving subject-matter experts, limit model training effectiveness. Privacy concerns are heightened due to edge devices' interaction with sensitive user data, making the affordability of data labeling and data privacy protection at the edge crucial issues. To tackle these challenges, we proposed Data Augmentation Random Padding to increase effective data for model training in CNNs, enhancing image classification accuracy. Despite limitations, we developed Semi-Supervised Learning (SSL) method \"AdaptMatch\" to utilize a large amount of unlabeled data and a small number of labeled data in centralized learning, improving learning speed and reducing label bias. To address the inadequacies of centralized data learning, we introduced a decentralized SSL method, Federated Incremental Learning (FedIL), for learning on edge devices while protecting privacy. However, FedIL's lower training efficiency and struggles with data imbalance led to the development of Federated Masked Autoencoder (FedMAE), which is also a decentralized semi-supervised learning method based on self-supervised learning. FedMAE enables asynchronous training of large-scale unlabeled images in federated learning, outperforming existing methods in handling highly imbalanced data.", "source_url": "https:\/\/openalex.org\/W7126832066", "pdf_url": "", "doi": "", "handle": "", "repository_name": "The Sydney eScholarship Repository (The University of Sydney)", "source_record_id": "W7126832066", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 586, "title": "Semi-Supervised Image Segmentation of Medical Data", "title_ar": "", "title_en": "Semi-Supervised Image Segmentation of Medical Data", "author": "Ove Nicolai Dalheim", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Norway", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "medical-ai", "research_subdomain": "image-segmentation", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; transfer-learning; self-semi-supervised; data-mining; cnn; image-segmentation", "keywords": "", "abstract": "Bladder cancer is the fourth most common cancer type in Norway, and tenth most common on a global scale. More and more tissue samples are sent to pathologists labs, increasing the workload and affecting the waiting time for patients. The corresponding increase is not seen in number of pathologists. Digitization and scanning of the tissue samples unveil the world of computational pathology, along with the many possibilities within it. Supervised approaches have been proposed within deep learning earlier, however, many express the lack of labeled data as a source for performance degradation. Convolutional neural networks have proven effective on image processing within the field of medicine. Some work based on semi-supervised approaches within computational pathology has been published in the recent years, however, most researchers are exploring supervised methods. This thesis proposes several methods to tile-wise segment histological images of bladder cancer into six different classes background, blood, damaged tissue, muscle tissue, stroma tissue and urothelium. A multiscale model is fed the tile at three different levels of magnification, and inherits capabilities from the VGG16 network through transfer learning. The proposed methods are either based on semi-supervised learning utilizing probability or clustering within a self-training process, or based on a combination of both expert and non-expert annotations. The dataset used in this thesis consists of about 360 whole-slide images, of which only 37 contain regions annotated by a pathologist, allowing for only about 10 % of the dataset to be utilized for supervised methods. Through the course of this thesis, a total of 145 new whole-slide images were introduced during training through semi-supervised learning, utilizing about 50 % of the dataset through various methods. The non-expert approach increased the F1-score for the muscle tissue class with 9.18 % from an initial 79.42 %, and the cluster-baser approach saw an increase of 1.38 % in accuracy from the initial 94.61 %. The method involving non-expert annotations outperformed both semi-supervised techniques with regards to segmentation, as a semi-supervised method will introduce new uncertain features to a deep learning model. This will have a strong impact on sensitive classes that are poorly represented in the training dataset. That said, a significant improvement is seen by using semi-supervised techniques as well, and scored better than non-expert with regards to classification.", "source_url": "https:\/\/openalex.org\/W3087951694", "pdf_url": "", "doi": "", "handle": "", "repository_name": "NORA - Norwegian Open Research Archives", "source_record_id": "W3087951694", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 588, "title": "Utilizing unlabeled data in cell type identification : A semi-supervised learning approach to classification", "title_ar": "", "title_en": "Utilizing unlabeled data in cell type identification : A semi-supervised learning approach to classification", "author": "Thijs Quast", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Linkoping University", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "machine-learning", "research_subdomain": "self-semi-supervised", "categories": "machine-learning; self-semi-supervised", "keywords": "", "abstract": "Recent research in bioinformatics has presented multiple cell type identification meth- dologies using single cell RNA sequence data (scRNA-seq). However, a consensus on which cell typing methodology consistently demonstrates superior performance remains absent. Additionally, very few studies approach cell type identification through a semi- supervised learning study, whereby the information in unlabeled data is leveraged to train an enhanced classifier. This paper presents cell annotation methodologies through self- learning and graph-based semi-supervised learning, in both raw count scRNA-seq data as well as in a latent embedding. I find that a self-learning framework enhances perfor- mance compared to a solely supervised learning classifier. Additionally, modelling on the latent data representations consistently outperforms modelling on the original data. The results show an overall accuracy of 96.12%, whereas additional models achieve an average precision rate of 95.12% and an average recall rate of 94.40%. The semi-supervised learn- ing approaches in this thesis compare favourable to scANVI in terms of accuracy, average precision rate, average recall rate and average f1-score. Moreover, results for alternative scenarios, in which cell types among training and test data do not perfectly overlap, are reported in this thesis.", "source_url": "https:\/\/openalex.org\/W3046627445", "pdf_url": "", "doi": "", "handle": "", "repository_name": "DiVA (Linkoping University)", "source_record_id": "W3046627445", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 591, "title": "Generative modelling under epistemic uncertainty", "title_ar": "", "title_en": "Generative modelling under epistemic uncertainty", "author": "Muhammad Mubashar", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Oxford Brookes University", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; generative-ai; robotics; predictive-analytics; llm; gan; diffusion", "keywords": "", "abstract": "The deployment of Deep Learning in safety-critical domains is hindered by pathological overconfidence and an inability to distinguish between aleatoric uncertainty (data ambiguity) and epistemic uncertainty (lack of knowledge). This thesis addresses these limitations by establishing a rigorous framework for Random-Set Deep Learning, shifting from point-estimate probabilities to belief functions over the power set of outcomes. Enabled by a novel Budgeting strategy that ensures scalability, this framework allows models to explicitly represent ignorance. While we validate these principles through Random-Set Neural Networks (RS-NN) for classification and a Unified Evaluation Framework, the primary contribution of this work lies in re-imagining Generative AI under epistemic uncertainty. We introduce Random-Set Large Language Models (RS-LLMs), which predict belief functions over token sets to quantify second-order uncertainty, thereby providing a robust mechanism for hallucination detection. Furthermore, we propose Epistemic Generative Adversarial Networks (GANs) and Epistemic Diffusion Models. By modeling the uncertainty of the generation process itself via second-order distributions, these architectures significantly mitigate mode collapse and enhance sample diversity compared to standard baselines. Finally, we demonstrate the practical utility of these generative capabilities in Autonomous Driving. We introduce the ROAD-INTENT dataset and utilize Random-Set Vision Language Models (RS-VLMs) to automate the annotation of actor intent with calibrated uncertainty scores, paving the way for safer, self-aware AI systems.", "source_url": "https:\/\/openalex.org\/W7135169854", "pdf_url": "", "doi": "10.24384\/j47e-p788", "handle": "", "repository_name": "Radar (Oxford Brookes University)", "source_record_id": "W7135169854", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 592, "title": "Towards data efficient generative models", "title_ar": "", "title_en": "Towards data efficient generative models", "author": "Zhaoyu Zhang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Queen's University Belfast", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "computer-vision", "research_subdomain": "gan", "categories": "machine-learning; computer-vision; generative-ai; gan; diffusion", "keywords": "", "abstract": "The past decade has seen significant advancements in generative models. As one of the most fundamental machine learning and computer vision tasks, generative models aim to produce realistic-looking synthetic samples by learning from real samples. However, training generative models from scratch commonly requires large amounts of data, which can be costly, time-consuming, or even impractical to collect and clean for specific domains. To address this, data-efficient generative models, which are designed to train with limited data from scratch, have gained significant attention. In this thesis, we focus on the theory and algorithms for training data-efficient generative models from scratch using two widely used generative models, i.e., generative adversarial networks (GANs) and diffusion models (DMs). For GANs, we address the challenges in data-efficient training from three different theoretical perspectives: the unfair min-max game, the overfitting of discriminator ($D$) and the leaking of augmentations problem. For DMs, we address the data efficiency challenge by better minimizing the total denoising score matching error during training. Firstly, we apply information game theory to analyze the GANs' min-max game and unveil a novel perspective on GAN training: the min-max game in existing GANs training is unfair. Based on this, we present a novel GAN called Information Gap GAN (IGGAN) to enhance the min-max game in GANs training. Specifically, IGGAN comprises a generator ($G$) and two discriminators ($D_1$ and $D_2$), each applying distinct data augmentation (DA) techniques. The information gap caused by various DA methods between $D_1$ and $D_2$ can alter the information received by each player in the min-max game, leading to incomplete information for all three players—$G$, $D_1$, and $D_2$—in IGGAN. As a result, IGGAN outperforms other existing methods, achieving better Inception Scores and lower Fréchet Inception Distances. Secondly we introduce a straightforward yet effective approach named Dual Adaptive Noise Injection (DANI) to mitigate the overfitting of $D$ in DE-GANs training. Specifically, DANI incorporates two adaptive strategies: adaptive injection probability and adaptive noise strength. By dynamically adjusting these factors, DANI effectively alleviates the overfitting of $D$. The strong performance and generalization ability of DANI demonstrate its effectiveness in improving GAN training under limited data conditions. Thirdly, we first illustrate that applying Data augmentation (DA) in DE-GANs can introduce out-of-distribution samples, leading to the undesirable leaking of augmentation issue in DE-GANs training. To address this, we present a straightforward yet powerful technique named Adaptive Negative Data Augmentation (ANDA) for DE-GANs. Specifically, ANDA enhances the augmented distribution of generated data by incorporating the augmented distribution of negative real data, which is generated by applying Negative Data Augmentation (NDA) to real data. This approach allows potential leaking samples to be presented as ``fake” instances to the discriminator in an adaptive manner, preventing the generator ($G$) from learning these samples and improving overall performance. We demonstrate that ANDA is effective in mitigating the leaking of augmentation issue. Finally, we focus on training diffusion models with limited data. We present a novel theoretical insight for diffusion models that two factors, i.e., the denoiser function hypothesis space and the number of training samples, can affect the denoising score matching error of all training samples. Based on this theoretical insight, it is clearly evident that the total denoising score matching error is hard to be minimized within the denoiser function hypothesis space of existing methods when training diffusion models with limited data. To address this, we design a new diffusion model with limited data called Limited Data Diffusion (LD-Diffusion), which consists of two main components: a compressing model and a novel mixed augmentation with fixed probability (MAFP) strategy. Comprehensive experiments show that LD-Diffusion greatly enhances the training of diffusion models with limited data, achieving state-of-the-art performance compared to other diffusion models.", "source_url": "https:\/\/openalex.org\/W7215036991", "pdf_url": "", "doi": "10.17034\/32826059", "handle": "", "repository_name": "Research Portal (Queen's University Belfast)", "source_record_id": "W7215036991", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 594, "title": "Generative diffusion models for wireless communication networks: detection, recovery, and decision-making", "title_ar": "", "title_en": "Generative diffusion models for wireless communication networks: detection, recovery, and decision-making", "author": "Changyuan Zhao", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Nanyang Technological University", "university_source": "repository_host", "college": "", "department": "", "country": "Singapore", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "computer-vision", "research_subdomain": "diffusion", "categories": "machine-learning; computer-vision; generative-ai; robotics; ai-applications; predictive-analytics; diffusion; ai-cybersecurity; ai-smart-cities", "keywords": "", "abstract": "Modern wireless communication systems are increasingly required to operate under uncertain and heterogeneous conditions, spanning signal inference under unknown noise distributions, reliable transmission in the presence of adversarial interference, and adaptive spectrum management in dynamic environments. While classical approaches have been developed for each of these settings individually, they typically rely on simplifying statistical assumptions that become harder to satisfy as deployment environments grow more complex. This thesis proposes a unified perspective: these three challenges, despite arising at different layers of the communication stack, are all instances of the same fundamental problem of conditional generation, namely, learning and sampling from the posterior distribution p(x \\ y) of a wireless variable given an observation. Under this view, a generative model captures prior knowledge about the signal or environment, the wireless system specifies the likelihood, and their combination through the posterior score drives inference, recovery, and control within a single probabilistic framework. Three contributions instantiate this perspective at progressively larger operational scales, each showing how conditional generation addresses limitations that are difficult to handle using purely model-based or discriminative approaches. In Chapter 3, the conditional generation perspective is applied at the signal level, addressing regression, classification, and multiple-input multiple-output (MIMO) detection tasks where the noise distribution is unknown or non-Gaussian. This chapter replaces direct prediction with posterior-based inference, enabling the diffusion mechanism to act as a universal scoring tool rather than a sampler. The Supervised Score-Based Model (SSM) reinterprets score-based inference as a noise-free gradient descent process, establishing a formal equivalence with gradient boosting and enabling deterministic, stable predictions without the stochastic overhead of generative models. The Maximum-Likelihood Estimation Inference (MLEI) framework repurposes the diffusion evidence lower bound as a distribution-free likelihood proxy, enabling maximum-likelihood-equivalent MIMO detection without any parametric assumption on the noise. Together, these contributions demonstrate that the diffusion mechanism extends naturally beyond generation to serve as a principled inference engine. In Chapter 4, the scale advances to the physical layer, addressing wireless image transmission under jamming and spoofing attacks. This chapter reformulates physical-layer security as a posterior sampling problem under a corrupted observation model, where the attack structure is encoded directly into the likelihood term. The SecDiff framework decomposes adversarial decoding into two attack-specific likelihoods: a masked inpainting constraint for subcarrier jamming, and an expectation-maximisation (EM) channel refinement loop for pilot spoofing. This enables plug-and-play adversarial recovery without retraining the underlying codec and without prior knowledge of the attack, establishing that the conditional generation viewpoint remains effective under structured, non-stochastic perturbations. In Chapter 5, the conditional generation perspective reaches the network level, addressing unmanned aerial vehicle (UAV)-based spectrum sensing and trajectory planning in dynamic, partially observable environments. This chapter reformulates spectrum cartography and multi-agent trajectory optimisation as two coupled conditional generation problems over complementary target distributions: the full radio power map given sparse sensor observations, and the optimal UAV action given the inferred spectrum state. A masked autoencoder reconstructor and a conditional diffusion policy are jointly optimised through the reconstruction loss, enabling perception and control to reinforce each other within a single generative formalism. This establishes that the conditional generation principle can scale from point-wise signal inference to multi-agent policy generation. In summary, the three chapters support a unified perspective that connects task-specific model-based and discriminative designs with posterior-driven generative intelligence. By reformulating bespoke estimators, classifiers, and controllers as task-specific conditional generation problems, this thesis provides a scalable and principled foundation for wireless intelligence. Building on this foundation, Chapter 6 identifies future research directions in which the same conditional generation principle can be extended further, including richer environment modelling for cognitive wireless agents, more energy-efficient posterior sampling for latency-constrained deployments, and tighter integration of spectrum perception and decision-making in next-generation low-altitude networks.", "source_url": "https:\/\/openalex.org\/W7212335836", "pdf_url": "", "doi": "10.32657\/10356\/221595", "handle": "", "repository_name": "DR-NTU (Nanyang Technological University)", "source_record_id": "W7212335836", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 597, "title": "Image restoration and representation with deep generative models", "title_ar": "", "title_en": "Image restoration and representation with deep generative models", "author": "Yufei Wang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Nanyang Technological University", "university_source": "repository_host", "college": "", "department": "", "country": "Singapore", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "computer-vision", "research_subdomain": "image-segmentation", "categories": "computer-vision; generative-ai; responsible-ai; ai-applications; image-classification; image-segmentation; gan; diffusion; fairness; ai-industry", "keywords": "", "abstract": "A fundamental challenge in the realm of computer vision lies in accurately modeling\/characterizing image distributions. For example, in high-level vision tasks, better representations of images in the latent space can significantly enhance downstream processes such as image classification and segmentation. Similarly, in image restoration, a more accurate model for the distributions of clean images conditioned on degraded images can yield results with better perceptual quality. However, modeling these distributions is highly challenging due to the high dimensionality of both images and their latent codes. In a relatively parallel research avenue, deep generative models have made remarkable strides, aiming to implicitly or explicitly model image distributions. These models offer potent tools for handling intricate distributions. However, directly applying these models is unfeasible and leads to suboptimal performances due to different domain priors and task formulations. The challenge of seamlessly and effectively integrating generative models into image restoration and representation tasks, along with incorporating task-specific priors, remains an open area for further exploration. This thesis focuses on exploring the potential of deep generative models in image restoration and representation tasks, spanning from high-level vision tasks, i.e., image classification, to low-level vision tasks, i.e., image compression and restoration. Firstly, for the image classification task, we introduce a novel variational inference framework to obtain a latent representation with enhanced generalization ability. We implicitly model the posterior distributions of images given their latent codes using a generative adversarial network to disentangle domain-invariant features from the provided training data. Subsequently, instead of implicitly modeling the posterior distributions, we explore the possibility of explicitly modeling them. Utilizing the power of Normalizing Flow, a generative model capable of obtaining the exact likelihood of a given sample, we apply it to the task of low-light image enhancement (LLIE), considering its inductive bias. While the flow-based model achieves promising results, the performance is constrained by the intrinsic limitations in the design of normalizing flow. Therefore, we delve into seamlessly integrating the prior knowledge of the LLIE task in raw image space into a diffusion framework to overcome the limitations in model designs. Although our diffusion-based solution has promising results, it still has two main limitations: first, even with the unique advantages of enhancement in raw space, the significant storage overhead of raw images hampers its application. Consequently, in the subsequent two works, we explore joint compression by accurately modeling the latent distributions of images and leveraging the capabilities of autoregressive models to further enhance coding efficiency. Second, while the proposed work shortens the inference path from tens or hundreds to only three steps compared with commonly used diffusion models, it still requires iterative evaluations and causes inference overhead. In the last work, we propose an acceleration strategy incorporating a distillation strategy with a novel self-consistency loss. We achieve SOTA performance in only one inference step in the super-resolution task, where the conditional distribution is relatively more complex. In summary, the thesis makes three primary contributions. First, we showcase the effectiveness and unique advantages of generative-based image restoration and representation methods. Second, we propose diverse approaches to seamlessly integrate the capabilities of deep generative models with the domain knowledge specific to image restoration and representation tasks. To validate the effectiveness of these proposed methods, extensive experiments are conducted across multiple datasets. The experimental results unequivocally demonstrate that our methods outperform prior state-of-the-art models. The efforts and achievements presented in this thesis not only underscore the practical capabilities of image restoration and representation techniques but also provide fundamental support for future research and applications within the industry.", "source_url": "https:\/\/openalex.org\/W4408105674", "pdf_url": "", "doi": "10.32657\/10356\/182543", "handle": "", "repository_name": "DR-NTU (Nanyang Technological University)", "source_record_id": "W4408105674", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 603, "title": "Generative Adversarial Networks for Classic Cryptanalysis", "title_ar": "", "title_en": "Generative Adversarial Networks for Classic Cryptanalysis", "author": "Deanne Charan", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "San Jose State University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "generative-ai", "research_subdomain": "gan", "categories": "machine-learning; deep-learning; generative-ai; gan", "keywords": "", "abstract": "The necessity of protecting critical information has been understood for millennia. Although classic ciphers have inherent weaknesses in comparison to modern ciphers, many classic ciphers are extremely challenging to break in practice. Machine learning techniques, such as hidden Markov models (HMM), have recently been applied with success to various classic cryptanalysis problems. In this research, we consider the effectiveness of the deep learning technique CipherGAN---which is based on the well- established generative adversarial network (GAN) architecture---for classic cipher cryptanalysis. We experiment extensively with CipherGAN on a number of classic ciphers, and we compare our results to those obtained using HMMs.", "source_url": "https:\/\/openalex.org\/W4206278187", "pdf_url": "https:\/\/scholarworks.sjsu.edu\/cgi\/viewcontent.cgi?article=2034&context=etd_projects", "doi": "10.31979\/etd.h2mh-uh52", "handle": "", "repository_name": "San José State University ScholarWorks (San Jose State University)", "source_record_id": "W4206278187", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 604, "title": "Bildgenerierung mittels Generative Adversarial Networks", "title_ar": "", "title_en": "", "author": "Meline A. Ayvazyan", "supervisor": "", "degree_type": "master", "degree_evidence": "نص السجل: «Masterarbeit»", "university": "Saxon State and University Library Dresden", "university_source": "repository_host", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2023, "language": "de", "research_domain": "generative-ai", "research_subdomain": "gan", "categories": "generative-ai; gan", "keywords": "", "abstract": "Das Hauptziel der Masterarbeit ist es, die Funktionsweise und das Verhalten von GANs in der Bildgenerierung zu untersuchen und dies anhand eines praktischen Beispiels zu verdeutlichen. Die Arbeit soll einen aktuellen Forschungsüberblick über die generative Modellierung, insbesondere im Bereich der Bildgenerierung, liefern und gleichzeitig die fachlichen Inhalte im Hinblick auf die Funktionsweise von GANs umfassend behandeln. Dabei werden sowohl theoretische Grundlagen als auch die praktische Umsetzung von GANs beleuchtet, um ein ganzheitliches Verständnis für dieses spannende Forschungsfeld zu vermitteln.", "source_url": "https:\/\/openalex.org\/W7139216419", "pdf_url": "https:\/\/tubaf.qucosa.de\/api\/qucosa%3A103059\/attachment\/ATT-0\/", "doi": "", "handle": "", "repository_name": "Qucosa (Saxon State and University Library Dresden)", "source_record_id": "W7139216419", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 610, "title": "Improve the efficiency of conditional generative models", "title_ar": "", "title_en": "Improve the efficiency of conditional generative models", "author": "Yanwu Xu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Boston University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; computer-vision; generative-ai; medical-ai; transfer-learning; video-analysis; gan; diffusion; medical-imaging", "keywords": "", "abstract": "Deep generative models have undergone significant advancements, enabling the production of high-fidelity data across various fields, including computer vision and medical imaging. The availability of paired annotations facilitates a controllable generative process through conditional generative models, which capture the conditional distribution P(X|Y), where X represents high-dimensional data and Y denotes the associated annotation. This controllability makes conditional generative models more preferable than ordinary generative models, which only model P(X). For instance, the latest generative AI techniques within the Artificial Intelligence Generated Content (AIGC) realm have unlocked the potential for flexible image and video generation\/editing based on text descriptions or “prompts.” Additionally, generative AI has enhanced model efficiency by supplementing datasets with synthesized data in scenarios where annotations are unavailable or imprecise. Despite these capabilities, challenges persist in ensuring efficient training of conditional generative models. These challenges include: 1) capturing the intricate relationship between data and its annotations, which can introduce biases; 2) higher computational resource requirements during training and inference compared to discriminative models; and 3) the need to balance speed and quality at inference time. To address these challenges, this thesis introduces four models aimed at enhancing the training and inference efficiency of conditional generative models without compromising quality. The first method focuses on conditional Generative Adversarial Networks (cGANs), proposing a novel training objective to improve stability and diversity in synthetic data generation. The second method involves a hybrid generative model that combines GANs and Diffusion-based models to alleviate the unstable training of GANs and accelerate the denoising process. The third model introduces a fine-tuning framework that utilizes pre-trained diffusion parameters for high-fidelity, fast sampling, and quick adaptation during training. The final method presents a super-efficient 3D diffusion model for high-fidelity 3D CT synthesis, addressing the efficiency and quality gap in current models. These methods collectively tackle the efficiency of generative models and enhance generative quality in both computer vision and medical domains, suggesting sustainable solutions for the future of generative AI.", "source_url": "https:\/\/openalex.org\/W7126296160", "pdf_url": "", "doi": "", "handle": "", "repository_name": "OpenBU (Boston University)", "source_record_id": "W7126296160", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 613, "title": "Dual-Attention Generative Adversarial Network and Flame and Smoke Analysis", "title_ar": "", "title_en": "Dual-Attention Generative Adversarial Network and Flame and Smoke Analysis", "author": "Yuchuan Li", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Ottawa", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; generative-ai; expert-systems; predictive-analytics; transformers; image-segmentation; video-analysis; gan", "keywords": "", "abstract": "Flame and smoke image processing and analysis could improve performance to detect smoke or fire and identify many complicated fire hazards, eventually to help firefighters to fight fires safely. Deep Learning applied to image processing has been prevailing in recent years among image-related research fields. Fire safety researchers also brought it into their studies due to its leading performance in image-related tasks and statistical analysis. From the perspective of input data type, traditional fire research is based on simple mathematical regressions or empirical correlations relying on sensor data, such as temperature. However, data from advanced vision devices or sensors can be analyzed by applying deep learning beyond auxiliary methods in data processing and analysis. Deep Learning has a bigger capacity in non-linear problems, especially in high-dimensional spaces, such as flame and smoke image processing. We propose a video-based real-time smoke and flame analysis system with deep learning networks and fire safety knowledge. It takes videos of fire as input and produces analysis and prediction for flashover of fire. Our system consists of four modules. The Color2IR Conversion module is made by deep neural networks to convert RGB video frames into InfraRed (IR) frames, which could provide important thermal information of fire. Thermal information is critically important for fire hazard detection. For example, 600 °C marks the start of a flashover. As RGB cameras cannot capture thermal information, we propose an image conversion module from RGB to IR images. The core of this conversion is a new network that we innovatively proposed: Dual-Attention Generative Adversarial Network (DAGAN), and it is trained using a pair of RGB and IR images. Next, Video Semantic Segmentation Module helps extract flame and smoke areas from the scene in the RGB video frames. We innovated to use synthetic RGB video data generated and captured from 3D modeling software for data augmentation. After that, a Video Prediction Module takes the RGB video frames and IR frames as input and produces predictions of the subsequent frames of their scenes. Finally, a Fire Knowledge Analysis Module predicts if flashover is coming or not, based on fire knowledge criteria such as thermal information extracted from IR images, temperature increase rate, the flashover occurrence temperature, and increase rate of lowest temperature. For our contributions and innovations, we introduce a novel network, DAGAN, by applying foreground and background attention mechanisms in the image conversion module to help reduce the hardware device requirement for flashover prediction. Besides, we also make use of combination of thermal information from IR images and segmentation information from RGB images in our system for flame and smoke analysis. We also apply a hybrid design of deep neural networks and a knowledge-based system to achieve high accuracy. Moreover, data augmentation is also applied on the Video Semantic Segmentation Module by introducing synthetic video data for training. The test results of flashover prediction show that our system has leading places quantitative and qualitative in terms of various metrics compared with other existing approaches. It can give a flashover prediction as early as 51 seconds with 94.5% accuracy before it happens.", "source_url": "https:\/\/openalex.org\/W3204711149", "pdf_url": "", "doi": "10.20381\/ruor-26991", "handle": "", "repository_name": "uO Research (University of Ottawa)", "source_record_id": "W3204711149", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 619, "title": "Nonparametric General Reinforcement Learning", "title_ar": "", "title_en": "Nonparametric General Reinforcement Learning", "author": "Jan Leike", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2016, "language": "en", "research_domain": "reinforcement-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; reinforcement-learning; predictive-analytics", "keywords": "", "abstract": "Reinforcement learning problems are often phrased in terms of Markov decision processes (MDPs). In this thesis we go beyond MDPs and consider reinforcement learning in environments that are non-Markovian, non-ergodic and only partially observable. Our focus is not on practical algorithms, but rather on the fundamental underlying problems: How do we balance exploration and exploitation? How do we explore optimally? When is an agent optimal? We follow the nonparametric realizable paradigm: we assume the data is drawn from an unknown source that belongs to a known countable class of candidates. First, we consider the passive (sequence prediction) setting, learning from data that is not independent and identically distributed. We collect results from artificial intelligence, algorithmic information theory, and game theory and put them in a reinforcement learning context: they demonstrate how an agent can learn the value of its own policy. Next, we establish negative results on Bayesian reinforcement learning agents, in particular AIXI. We show that unlucky or adversarial choices of the prior cause the agent to misbehave drastically. Therefore Legg-Hutter intelligence and balanced Pareto optimality, which depend crucially on the choice of the prior, are entirely subjective. Moreover, in the class of all computable environments every policy is Pareto optimal. This undermines all existing optimality properties for AIXI. However, there are Bayesian approaches to general reinforcement learning that satisfy objective optimality guarantees: We prove that Thompson sampling is asymptotically optimal in stochastic environments in the sense that its value converges to the value of the optimal policy. We connect asymptotic optimality to regret given a recoverability assumption on the environment that allows the agent to recover from mistakes. Hence Thompson sampling achieves sublinear regret in these environments. AIXI is known to be incomputable. We quantify this using the arithmetical hierarchy, and establish upper and corresponding lower bounds for incomputability. Further, we show that AIXI is not limit computable, thus cannot be approximated using finite computation. However there are limit computable ε-optimal approximations to AIXI. We also derive computability bounds for knowledge-seeking agents, and give a limit computable weakly asymptotically optimal reinforcement learning agent. Finally, our results culminate in a formal solution to the grain of truth problem: A Bayesian agent acting in a multi-agent environment learns to predict the other agents' policies if its prior assigns positive probability to them (the prior contains a grain of truth). We construct a large but limit computable class containing a grain of truth and show that agents based on Thompson sampling over this class converge to play ε-Nash equilibria in arbitrary unknown computable multi-agent environments.", "source_url": "https:\/\/openalex.org\/W2558819991", "pdf_url": "https:\/\/arxiv.org\/pdf\/1611.08944", "doi": "10.25911\/5d76346c2e2be", "handle": "", "repository_name": "arXiv (Cornell University)", "source_record_id": "W2558819991", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 620, "title": "Curiosity-based Reinforcement Learning", "title_ar": "", "title_en": "Curiosity-based Reinforcement Learning", "author": "GIANLUCA MANNONI", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Pisa", "university_source": "repository_host", "college": "", "department": "", "country": "Italy", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "computer-vision", "research_subdomain": "multimodal", "categories": "deep-learning; computer-vision; reinforcement-learning; multimodal", "keywords": "", "abstract": "The thesis initially focused on the study of the main Reinforcement Learning methods, highlighting their principal limitations and their temporal evolution. Subsequently, the sparse reward problem was analyzed in detail, which is associated with the lack of updates of the current policy due to the low frequency of feedback provided by the environment. To address this issue, several curiosity-based methods proposed in the literature were examined. Among these, two different methodologies were implemented and experimentally compared with the selected Reinforcement Learning algorithm in applications characterized by sparse rewards and increasing levels of complexity. Three different types of tasks were therefore considered: the control of an inverted pendulum, a manipulation task based on visual observations only, and a navigation task in the presence of obstacles, which combines visual observations with other types of sensory information. In the first experiment, it was observed that the integration of curiosity modules does not provide significant benefits, while incurring a higher computational cost. In the second experiment, instead, the contribution of the modules appears to be more evident. It was therefore concluded that curiosity-based methods are determinant for achieving a performance improvement in the specific application scenario. In the final experiment, it was observed that the introduction of multimodal observations led to a marked deterioration in the performance of one of the two methods, mainly due to the higher sensitivity of the module to visual observations. This behavior can be attributed to a component fixed at the beginning of the training process, which is unable to adapt the relative importance of the received observations. Conversely, the second method updates the internal feature extractor of the model, thereby avoiding this limitation and ensuring, also in this case, superior performance compared to the other evaluated methods.", "source_url": "https:\/\/openalex.org\/W7142563277", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Electronic Theses and Dissertations Repository (University of Pisa)", "source_record_id": "W7142563277", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 623, "title": "Learning Tetris with reinforcement learning", "title_ar": "", "title_en": "Learning Tetris with reinforcement learning", "author": "Ioannis Tsirovasilis ; Ιωάννης Τσιροβασίλης", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Piraeus", "university_source": "repository_host", "college": "", "department": "", "country": "Greece", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "computer-vision", "research_subdomain": "video-analysis", "categories": "machine-learning; computer-vision; reinforcement-learning; optimization; video-analysis", "keywords": "", "abstract": "The last decades, scientists have expressed an increasing interest for the game Tetris and more precisely for efficient algorithms that can score the most in-game points. A plethora of approaches have been tried out, including genetic algorithms, linear programming, cross-entropy and natural policy gradient, but none of them competes with experts players playing under no time pressure. In recent years, scientists have started applying reinforcement learning in Tetris as it displays effective results in adapting to video game environments, exploit mechanisms and deliver extreme performances. Current thesis aims to introduce Memory Based Learning, a reinforcement learning algorithm which uses a memory that helps in the training process by replaying past experiences.", "source_url": "https:\/\/openalex.org\/W3214945965", "pdf_url": "", "doi": "10.26267\/unipi_dione\/1314", "handle": "", "repository_name": "Dione (University of Piraeus)", "source_record_id": "W3214945965", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 631, "title": "Discovery in Reinforcement Learning", "title_ar": "", "title_en": "Discovery in Reinforcement Learning", "author": "Vivek Veeriah", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Michigan", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "computer-vision", "research_subdomain": "predictive-analytics", "categories": "machine-learning; deep-learning; computer-vision; reinforcement-learning; robotics; predictive-analytics", "keywords": "", "abstract": "This thesis focuses on Reinforcement Learning (RL) which considers an agent that makes sequen- tial decisions in an environment to maximize its reward. Recent breakthroughs on domains such as Go, Atari games, and Robotics are a result of combining advances from Deep Learning with RL algorithms. This combination of RL with deep neural networks has moved the field from hand- designing feature representations to learning features directly from raw data (e.g., pixel images). Despite these advancements, RL practitioners still need to define what predictive knowledge an agent needs to learn, instead of the agent autonomously discovering such knowledge. Developing the ability to discover what predictive knowledge is useful to learn about could allow the RL agents to learn more efficiently. In this thesis, we identify some of the challenges towards designing agents with the aforementioned ability, and propose and evaluate methods to address them. In the first part of this thesis, we present an approach for discovering multiple subgoals to learn optimal policies from a single stream of experience. We show that this approach is a useful pre- training procedure and as a source of auxiliary task updates when there is a main task of interest. In the second part of this thesis, we present three approaches to discover useful knowledge of various forms to maximize the agent’s task performance. The first one introduces an architecture and an associated meta-gradient algorithm to discover predictive questions about the agent’s ex- perience to drive representation learning in RL agents. We show that it helps in faster learning when used as auxiliary tasks. The second work introduces another meta-gradient approach to dis- cover temporal-abstractions in the form of options that enables a hierarchical RL agent to learn faster on new, unseen tasks. The third one presents an algorithm to select a small number of affor- dances in the form of actions or options from a continuous space to improve the performance of a model-based planning agent on hierarchical tasks. The final part studies the problem of learning and maintaining accurate knowledge in the form of option-conditional predictions in a partially-observable environment. Specifically, we implement and empirically demonstrate the thought experiment by Ring [2021], which provided a detailed blueprint of how high-level, abstract knowledge could potentially be represented as layered pre- dictions of agent’s sensorimotor stream in navigation environments. In addition to this empirical demonstration, we introduce an approach to discover those option-conditional predictions which were previously hand-defined and demonstrate the feasibility of this approach.", "source_url": "https:\/\/openalex.org\/W4294885165", "pdf_url": "", "doi": "10.7302\/6337", "handle": "", "repository_name": "Deep Blue (University of Michigan)", "source_record_id": "W4294885165", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 636, "title": "Multimodal perception for autonomous driving", "title_ar": "", "title_en": "Multimodal perception for autonomous driving", "author": "Beltrán de la Cita, Jorge", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Carlos III University of Madrid", "university_source": "repository_host", "college": "", "department": "", "country": "Spain", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "deep-learning; computer-vision; robotics; ai-applications; multimodal; object-detection; ai-industry; ai-smart-cities", "keywords": "", "abstract": "Autonomous driving is set to play an important role among intelligent transportation systems in the coming decades. The advantages of its large-scale implementation –reduced accidents, shorter commuting times, or higher fuel efficiency– have made its development a priority for academia and industry. However, there is still a long way to go to achieve full self-driving vehicles, capable of dealing with any scenario without human intervention. To this end, advances in control, navigation and, especially, environment perception technologies are yet required. In particular, the detection of other road users that may interfere with the vehicle’s trajectory is a key element, since it allows to model the current traffic situation and, thus, to make decisions accordingly. The objective of this thesis is to provide solutions to some of the main challenges of on-board perception systems, such as extrinsic calibration of sensors, object detection, and deployment on real platforms. First, a calibration method for obtaining the relative transformation between pairs of sensors is introduced, eliminating the complex manual adjustment of these parameters. The algorithm makes use of an original calibration pattern and supports LiDARs, and monocular and stereo cameras. Second, different deep learning models for 3D object detection using LiDAR data in its bird’s eye view projection are presented. Through a novel encoding, the use of architectures tailored to image detection is proposed to process the 3D information of point clouds in real time. Furthermore, the effectiveness of using this projection together with image features is analyzed. Finally, a method to mitigate the accuracy drop of LiDARbased detection networks when deployed in ad-hoc configurations is introduced. For this purpose, the simulation of virtual signals mimicking the specifications of the desired real device is used to generate new annotated datasets that can be used to train the models. The performance of the proposed methods is evaluated against other existing alternatives using reference benchmarks in the field of computer vision (KITTI and nuScenes) and through experiments in open traffic with an automated vehicle. The results obtained demonstrate the relevance of the presented work and its suitability for commercial use.", "source_url": "https:\/\/openalex.org\/W2964186846", "pdf_url": "", "doi": "", "handle": "", "repository_name": "e-Archivo (Carlos III University of Madrid)", "source_record_id": "W2964186846", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 646, "title": "Hybrid Optimization and Learning-Based Techniques for Autonomous Driving", "title_ar": "", "title_en": "Hybrid Optimization and Learning-Based Techniques for Autonomous Driving", "author": "Kürsat Petek", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Universitätsbibliothek Freiburg", "university_source": "repository_host", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "computer-vision", "research_subdomain": "image-segmentation", "categories": "machine-learning; deep-learning; computer-vision; robotics; ai-applications; self-semi-supervised; predictive-analytics; multimodal; image-segmentation; video-analysis; ai-smart-cities", "keywords": "", "abstract": "Autonomous driving has progressed rapidly in recent years, driven largely by advances in machine learning, computer vision, and probabilistic optimization methods. Despite this progress, the deployment of autonomous systems in complex, safety-critical real- world settings such as urban driving, dynamic traffic scenes, or large-scale outdoor environments remains a major open challenge. Real-world autonomy requires systems that can calibrate and localize themselves reliably, reason about their surroundings with minimal human supervision, and continuously adapt to changing environments. This thesis develops a set of methods that jointly address these challenges by combining geometric optimization with learning-based perception, and by designing label-efficient and self-supervised systems that scale to real-world deployment. In the first part, we study hybrid optimization methods for calibration and localiza- tion, focusing on the tight coupling of classical geometry with learned perception cues. We introduce a fully automatic camera–LiDAR calibration framework that removes the need for targets, initial guesses, or manual intervention by jointly exploiting motion cues and deep 2D–3D correspondences. Building on this idea, we propose a sparse-map localization system that integrates uncertainty-aware semantic predictions into a pose graph, enabling robust operation even in highly occluded urban environments with minimal map information. The second part focuses on online semantic Bird’s-Eye-View (BEV) mapping, where the goal is to infer a semantic map of the surrounding scene directly from frontal-view images in real time. We first investigate multi-modal fusion through BEVCar, which combines cameras and radar to obtain long-range, all-weather BEV maps using low- cost sensor setups. We then reduce the reliance on BEV annotations with SkyEye, a self-supervised method that learns BEV semantics from monocular images while using only frontal-view semantic labels as supervision. Finally, LetsMap removes this remaining supervision by jointly learning geometry and scene representation from raw RGB videos alone, achieving state-of-the-art label-efficient BEV mapping with only a tiny fraction of annotated BEV data. The third part addresses the challenge that modern perception systems rely on thousands of densely annotated images which is a major bottleneck for deploying autonomy at scale. We study how to drastically reduce this dependence by developing label-efficient and continually adapting panoptic segmentation methods that can be trained from data obtained within hours rather than months. SPINO shows that combining foundation-model features with only a handful of labeled images is sufficient to produce high-quality panoptic pseudo-labels at scale. PASTEL extends this idea with normalized-cut–based instance separation and targeted self-training, achieving even stronger performance under strict annotation budgets. Finally, CoDEPS enables continual test-time learning for both depth and panoptic segmentation through a geometry-aware mixing strategy and replay-based stabilization, allowing models to adapt to new environments during deployment without forgetting previous ones. Across these contributions, the thesis shows that combining geometric structure with modern learning techniques enables perception systems that are more robust, more data-efficient, and more adaptable than traditional supervised approaches. We demonstrate these methods extensively on diverse real-world driving datasets and robotic platforms, highlighting their potential to bring scalable, reliable autonomy closer to widespread deployment.", "source_url": "https:\/\/openalex.org\/W7196982350", "pdf_url": "", "doi": "10.6094\/unifr\/285382", "handle": "", "repository_name": "FreiDok plus (Universitätsbibliothek Freiburg)", "source_record_id": "W7196982350", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 647, "title": "Interaction-Aware Maneuver and Lane Change Prediction Using Deep Learning Architectures for Autonomous Driving", "title_ar": "", "title_en": "Interaction-Aware Maneuver and Lane Change Prediction Using Deep Learning Architectures for Autonomous Driving", "author": "Petrit Rama", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Kaiserslautern University of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; robotics; responsible-ai; ai-applications; predictive-analytics; rnn; transformers; gnn; multimodal; xai; ai-smart-cities", "keywords": "", "abstract": "Maneuver prediction is a critical component of decision-making in autonomous driving, particularly in complex urban environments where traffic participants continuously interact under uncertainty. Accurate and interpretable prediction of future driving maneuvers, such as lane changes, is essential for safe and anticipatory planning. This thesis investigates interaction-aware maneuver prediction from the ego-vehicle perspective using deep learning-based approaches that integrate spatial reasoning, temporal modeling, and multi-modal perception. The research presents a progression of architectures that address maneuver and lane change prediction at increasing levels of complexity. First, interaction-aware models based on graph neural networks (GNNs) and recurrent neural networks (RNNs) jointly capture spatial interactions between traffic agents and the temporal evolution of driving behavior for scene understanding. Building on this, a cognitively inspired two-layer hierarchical decision-making framework separates planned maneuvers from stimuli and cause reasoning to improve interpretability. Subsequently, a deep multi-modal architecture based on Edge-Featured Graph Attention Networks (EGATs) is developed for lane change prediction, integrating visual perception, spatial interaction graphs, ego-vehicle state signals, and lane-level information. To support this research, a new in-house multi-modal dataset, WylonSet, is introduced and later extended to WylonSet++, providing high-quality annotations for lane change maneuvers and, in its extended version, for maneuver phases in diverse urban traffic scenarios. Finally, a transformer-based multi-modal architecture is proposed for multi-horizon lane change and phase prediction. The model employs attention-based fusion across modalities and incorporates uncertainty estimation at the modality, fusion, and prediction levels, enabling interpretable and future-aware behavior prediction. Experiments on four real-world driving datasets show that interaction-aware, multi-modal, and attention-based architectures consistently improve prediction accuracy and temporal consistency compared with uni-modal and non-interaction-aware variants. The transformer-based architecture predicts lane change maneuvers and their phases across horizons of up to 3 s, achieving an overall macro F1-score above 97% on WylonSet++ and detecting lane change events on average about 1 s before their onset. Overall, this thesis contributes new models, datasets, and evaluation methodology for maneuver and lane change prediction, providing a foundation for interpretable and deployment-oriented decision-making systems in autonomous driving.", "source_url": "https:\/\/openalex.org\/W7222983659", "pdf_url": "", "doi": "10.26204\/kluedo\/13713", "handle": "", "repository_name": "Publication Server of Kaiserslautern University of Technology (Kaiserslautern University of Technology)", "source_record_id": "W7222983659", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 649, "title": "Static and Dynamic Affordance Learning in Vision-based Direct Perception for Autonomous Driving", "title_ar": "", "title_en": "Static and Dynamic Affordance Learning in Vision-based Direct Perception for Autonomous Driving", "author": "Jean Marie Uwabeza Vianney", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Waterloo", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "predictive-analytics", "categories": "machine-learning; deep-learning; computer-vision; robotics; ai-applications; predictive-analytics; ai-smart-cities", "keywords": "", "abstract": "The recent development in autonomous driving involves high-level computer vision and detailed road scene understanding. Today, most autonomous vehicles are using the mediated perception approach for path planning and control, which highly rely on high-definition 3D maps and real-time sensors. Recent research efforts aim to substitute the massive HD maps with coarse road attributes. In this thesis, We follow the direct perception-based method to train a deep neural network for affordance learning in autonomous driving. The goal and the main contributions of this thesis are in two folds. Firstly, to develop the affordance learning model based on freely available Google Street View panoramas and Open Street Map road vector attributes. Driving scene understanding can be achieved by learning affordances from the images captured by car-mounted cameras. Such scene understanding by learning affordances may be useful for corroborating base-maps such as HD maps so that the required data storage space is minimized and available for processing in real-time. We compare capability in road attribute identification between human volunteers and the trained model by experimental evaluation. The results indicate that this method could act as a cheaper way for training data collection in autonomous driving. The cross-validation results also indicate the effectiveness of the trained model. Secondly, We propose a scalable and affordable data collection framework named I2MAP (image-to-map annotation proximity algorithm) for autonomous driving systems. We built an automated labeling pipeline with both vehicle dynamics and static road attributes. The data collected and annotated under our framework is suitable for direct perception and end-to-end imitation learning. Our benchmark consists of 40,000 images with more than 40 affordance labels under various day time and weather even with very challenging heavy snow. We train and evaluate a ConvNet based traffic flow prediction model for driver warning and suggestion under low visibility condition.", "source_url": "https:\/\/openalex.org\/W3004650162", "pdf_url": "", "doi": "", "handle": "", "repository_name": "UWSpace (University of Waterloo)", "source_record_id": "W3004650162", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 650, "title": "Distributed Intelligence for 3D Object Tracking in Autonomous Vehicles, A Multi-Sensor Approach to Object Tracking and Planning for Self-Driving Cars, Real-Time 3D Perception and Path Planning in Autonomous Systems,,Vision-Based Localization and Distributed Architecture for Driverless Cars,A Hybrid Deep Learning Framework for Object Detection and Trajectory Planning in Autonomous Vehicles", "title_ar": "", "title_en": "Distributed Intelligence for 3D Object Tracking in Autonomous Vehicles, A Multi-Sensor Approach to Object Tracking and Planning for Self-Driving Cars, Real-Time 3D Perception and Path Planning in Autonomous Systems,,Vision-Based Localization and Distributed Architecture for Driverless Cars,A Hybrid Deep Learning Framework for Object Detection and Trajectory Planning in Autonomous Vehicles", "author": "Manas Madine", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Massachusetts Amherst", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; robotics; ai-applications; predictive-analytics; object-detection; medical-imaging; ai-industry; ai-smart-cities", "keywords": "", "abstract": "Abstract-I: Autonomous Navigating car : To make car move as fast as possible and with absolutely no collisions. And Controlling a car with high dimensional sensory inputs and Driving a car autonomously around a racetrack with obstacles like pedestrians and other vehicles and even traffic rules. We created a deep Q-network (DQN) agent to perform the task of autonomous car driving from raw sensory inputs. We evaluated our agent’s performance against several standard agents in a racing simulation environment (CARLA). Our results indicate that our DQN agent is capable of successfully controlling a car to navigate around a simulation environment(CARLA). Abstract-II: 3D Object Tracking in Autonomous Driving: we developed a stereo-camera based 3D multi Vehicle tracking System. The main objective of this system is to accurately predict location and orientation of vehicles from stereo camera data. It has three modules: a 2D object detection network, 3D position extraction and 3D object correlation. This system utilizes the Kalman filtering to improve the robustness of missed detections Abstract-III: trajectory planning: The planning module in an autonomous navigation can be considered as the system’s brain where the vehicle trajectory for the next seconds is planned. In this module, we make use of the perception data from the 3D object tracking to plan a secure and smooth trajectory for the vehicle. The trajectory planning objective is to provide the vehicle with a secure trajectory constrained by the vehicle dynamics limits, the navigation comfort and safety, and the traffic rules. Once the desired trajectory is generated, the next step is to control the vehicle in order to track this trajectory with the desired speed profile also defined by the planning module. The vehicle control is indeed the control of the vehicle actuators, such as the steering wheel, the accelerator and the braking force, in order to manage the vehicle motions. Abstract-IV: Distributed System Architecture and Development Process: For autonomous driving, complex autonomous driving algorithms, including perception, 3D object tracking, localization, planning, and control, are required with many heterogeneous sensors and computers. To manage the complexity of the driving algorithms and the heterogeneity of the system components, we applied distributed system architecture to the autonomous driving system, and a system platform for the distributed system of an autonomous car. The development process provides the guidelines to design and develop the distributed system of an autonomous vehicle. A time-triggered network protocol, FlexRay, to be applied as the main network of the software platform to improve the network bandwidth, fault tolerance, and system performance.", "source_url": "https:\/\/openalex.org\/W7135301869", "pdf_url": "", "doi": "", "handle": "", "repository_name": "ScholarWorks@UMassAmherst (University of Massachusetts Amherst)", "source_record_id": "W7135301869", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 651, "title": "Deep Structured Multi-Task Learning for Computer Vision in Autonomous Driving", "title_ar": "", "title_en": "Deep Structured Multi-Task Learning for Computer Vision in Autonomous Driving", "author": "Marvin Teichmann", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Cambridge", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "image-segmentation", "categories": "machine-learning; deep-learning; computer-vision; robotics; responsible-ai; predictive-analytics; cnn; image-segmentation; xai", "keywords": "", "abstract": "The field of computer vision is currently dominated by deep learning advances. Convolutional Neural Networks (CNNs) have become the predominant tool for solving almost any computer vision task, so state-of-the-art systems have been built by using the predictive capabilities of Convolutional Neural Networks (CNNs). Many of those systems use simple encoder–decoder based design, where an off-the-shelf CNN architecture is combined with a task-specific decoder and loss function in order to create an end-to-end trainable model. This ultimately raises the question of whether these kinds of models are the future of computer vision. In this thesis we argue that this is not the case. We start off by discussing three limitations of simple end-to-end training. We proceed by showing how it is possible to overcome those limitations by using an approach that we call structured modelling. The idea is to use CNNs to compute a rich semantic intermediate representation which is then used to solve the actual problem by applying a geometric and task-related structure. In this work we solve the localization, segmentation and landmark recognition task using structured modelling, and we show that this approach can improve generalization, interpretability and robustness. We also discuss how this approach is particularly useful for real-time applications such as autonomous driving. Visual perception is a multi-module problem that requires several different computer vision tasks to be solved. We discuss how, by sharing computations, we can improve not only the inference speed but also the prediction performance by using the structural relationship between the tasks. Lastly, we demonstrate that structured modelling is able to achieve state-of-the-art performance, making it a very relevant approach for solving current and future computer vision problems.", "source_url": "https:\/\/openalex.org\/W3049041115", "pdf_url": "", "doi": "10.17863\/cam.56207", "handle": "", "repository_name": "Apollo (University of Cambridge)", "source_record_id": "W3049041115", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 652, "title": "Learning to Drive: Exploiting Deep Models for Autonomous Driving", "title_ar": "", "title_en": "Learning to Drive: Exploiting Deep Models for Autonomous Driving", "author": "Wenjie Luo", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Toronto", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "computer-vision", "research_subdomain": "image-segmentation", "categories": "machine-learning; deep-learning; computer-vision; robotics; responsible-ai; predictive-analytics; cnn; image-segmentation; xai", "keywords": "", "abstract": "Building self-driving vehicles is exciting and promising. It is going to transform the way we live and provide safety, efficiency, and mobility for everyone. In this thesis, I present a collection of our work in the direction of building smart self-driving vehicles from understanding convolutional neural networks (CNNs) to the full autonomy stacks, including perception, prediction, and planning. First, we study the property of CNNs and provide theoretical analysis on the receptive field. We use Fourier transform and center limit theorem to show that the effective receptive field only occupies a fraction of the theoretical one, especially in deep models. This can provide insights for future CNNs design. Next, we push the state of the art for tasks in the autonomy stacks using deep CNNs. For depth estimation with stereo cameras, we develop a deep matching network that simplifies previous approaches by using a dot-product layer and incorporates a multi-class classification loss, allowing the network to calibrate scores among larger contexts. These greatly improve the runtime and achieve much better matching results. Furthermore, we extend the idea to optical flow where the matching is done in 2D space. Combined with an instance segmentation algorithm, we achieve state-of-the-art results on KITTI optical flow benchmark. Third, we propose new formulations for the self-driving system, i.e. using a joint model for 3D detection, prediction, and tracking. Our model takes a sequence of consecutive LiDAR sweeps and predicts the bounding boxes for vehicles at the current time step as well as one second into the future. Tracking is done by greedily checking the overlap between current detections and predictions from the past. This approach bridges the gap between perception and prediction to achieve better performance both in accuracy and runtime. Then, we take this one step further to construct a deep structured model for interpretable neural motion planning, where the CNN also predicts a 3D cost volume, expanding on time and x-y spatial dimensions. The final planning trajectory is generated with a sampling-based inference algorithm. We conduct offline testing and show that our approach is better than several baselines.", "source_url": "https:\/\/openalex.org\/W3154873709", "pdf_url": "", "doi": "", "handle": "", "repository_name": "TSpace (University of Toronto)", "source_record_id": "W3154873709", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 655, "title": "Learning autonomous driving policies for dynamic urban and racing environments", "title_ar": "", "title_en": "Learning autonomous driving policies for dynamic urban and racing environments", "author": "Peide Cai", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "computer-vision", "research_subdomain": "xai", "categories": "deep-learning; computer-vision; reinforcement-learning; robotics; responsible-ai; ai-applications; multimodal; xai; ai-smart-cities", "keywords": "", "abstract": "It is believed that self-driving vehicles will play an essential role in reducing road accidents, alleviating traffic congestion, and reducing energy consumption in transportation, thus providing safety, efficiency, and convenience for everyone. This thesis focuses on building autonomous driving systems leveraging deep learning under complex contexts from normal-speed urban to high-speed drifting\/racing scenarios. The first part studies the frameworks of end-to-end driving in dynamic urban scenarios based on imitation learning (IL).We progressively consider all-day and all-weather driving conditions using a combination of multi-modal sensors, i.e., cameras, lidars, and radars. The uncertainty measures are also utilized to indicate dangerous cases and improve the interpretability of the driving system. In the second part, we focus on local motion planning using semantic state abstractions rather than learning the entire navigation stack to achieve more generalizable and strategic driving performance in interaction-intensive environments, e.g., unregulated intersections\/ roundabouts. The interactions among different road users are modeled implicitly through graph attention networks (GAT), where the inter-vehicle influences are adaptively weighted according to specific contexts. By training the whole network through deep reinforcement learning (DRL), the proposed method performs better in balancing driving safety and efficiency than the baselines. The learned policy can also be transferred into a real-world dataset thanks to the decoupled perception-planning framework. Furthermore, the third part highlights highly dynamic drifting\/racing scenarios to facilitate agile autonomous driving near the handling limits under high-speed conditions. We first study the drift cornering control problems based on model-free DRL approaches, then combine IL and model-based DRL to solve a vision-based autonomous racing task. The proposed method is validated both in a realistic simulator and on a physical RC car. Related methodologies also support the 1st HKUST autonomous RC-car racing competition1.", "source_url": "https:\/\/openalex.org\/W4307183652", "pdf_url": "", "doi": "10.14711\/thesis-991013106451703412", "handle": "", "repository_name": "", "source_record_id": "W4307183652", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 656, "title": "Generating and Exploiting Deep Learning Variants to Increase Utilization of the Heterogeneous Resources in Autonomous Driving Platforms", "title_ar": "", "title_en": "Generating and Exploiting Deep Learning Variants to Increase Utilization of the Heterogeneous Resources in Autonomous Driving Platforms", "author": "Roger Pujol Torramorell", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "deep-learning; computer-vision; robotics; object-detection", "keywords": "", "abstract": "Nowadays, Deep learning-based solutions and, in particular, deep neural networks (DNNs) are getting into several core functionalities in critical real-time embedded systems (CRTES), like those in planes, cars, and satellites, from vision-based perception (object detection and object tracking) systems to trajectory planning. As a result, several deep learning instances are running simultaneously at any time on the same computing platform. However, while modern computing platforms offer a variety of computing elements (e.g., CPUs, GPUs, and specific accelerators) in which those DNN instances can be executed depending on their computational requirements and temporal constraints. Currently, most DNNs are mainly programmed to exploit one particular computing element, regular cores of the GPUs. This lack of variety causes a resource imbalance and under-utilization of the various computing element resources when executing several DNN instances, causing an increase in DNN tasks' execution time requirements. In this Thesis, (a) we develop different variants (implementation) of well-known DNN libraries used in the Apollo Autonomous Driving software for each of the computing elements of the latest NVIDIA Xavier system-on-chip. Each variant is configured to balance resource requirements and performance: the regular CPU core implementation that can run on 2, 4, and 6 cores (always leaving 2 cores free for other computations); the GPU with regular and Tensor cores variants that can run on 4 or 8 GPU's Stream Multiprocessors (SM); and 1 or 2 NVIDIA's Deep Learning Accelerators (NVDLA); (b) we show that each particular variant\/configuration offers different resource utilization\/performance point. (c) we show how those heterogeneous computing elements can be exploited by a static scheduler to sustain the execution of multiple and diverse DNN variants on the same platform.", "source_url": "https:\/\/openalex.org\/W3121071986", "pdf_url": "", "doi": "", "handle": "", "repository_name": "LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas)", "source_record_id": "W3121071986", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 658, "title": "Deep End-to-end Network for 3D Object Detection in the Context of Autonomous Driving", "title_ar": "", "title_en": "Deep End-to-end Network for 3D Object Detection in the Context of Autonomous Driving", "author": "Dominik Jargot", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Delft University of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Netherlands", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "deep-learning; computer-vision; robotics; object-detection", "keywords": "", "abstract": "Nowadays, autonomous driving is a trending topic in the automotive field. One of the most crucial challenges of autonomous driving research is environment perception. Currently, many techniques achieve satisfactory performance in 2D object detection using camera images. Nevertheless, such 2D object detection might be not sufficient for autonomous driving applications as the vehicle is operating in a 3D world where all the dimensions have to be considered. In this thesis a new method for 3D object detection, using deep learning approach is presented. The proposed architecture is able to detect cars using data from images and point clouds. The proposed network does not use any hand-crafted features and is trained in an end-to-end manner. The network is trained and evaluated with the widely used KITTI dataset. The proposed method achieves an average precision of 81.38%, 67.02%, and 65.30% on the easy, moderate, and hard subsets of the KITTI validation dataset, respectively. The average inference time per scene is 0.2 seconds.", "source_url": "https:\/\/openalex.org\/W3038194063", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Research Repository (Delft University of Technology)", "source_record_id": "W3038194063", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 660, "title": "Lidar-based scene understanding for autonomous driving using deep learning", "title_ar": "", "title_en": "Lidar-based scene understanding for autonomous driving using deep learning", "author": "Víctor Vaquero Gómez", "supervisor": "", "degree_type": "phd", "degree_evidence": "نص السجل: «PhD»", "university": "Universitat Politècnica de Catalunya", "university_source": "repository_host", "college": "", "department": "", "country": "Spain", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "image-segmentation", "categories": "machine-learning; deep-learning; computer-vision; robotics; ai-applications; predictive-analytics; image-segmentation; ai-smart-cities", "keywords": "", "abstract": "With over 1.35 million fatalities related to traffic accidents worldwide, autonomous driving was foreseen at the beginning of this century as a feasible solution to improve security in our roads. Nevertheless, it is meant to disrupt our transportation paradigm, allowing to reduce congestion, pollution, and costs, while increasing the accessibility, efficiency, and reliability of the transportation for both people and goods. Although some advances have gradually been transferred into commercial vehicles in the way of Advanced Driving Assistance Systems (ADAS) such as adaptive cruise control, blind spot detection or automatic parking, however, the technology is far from mature. A full understanding of the scene is actually needed so that allowing the vehicles to be aware of the surroundings, knowing the existing elements of the scene, as well as their motion, intentions and interactions. In this PhD dissertation, we explore new approaches for understanding driving scenes from 3D LiDAR point clouds by using Deep Learning methods. To this end, in Part I we analyze the scene from a static perspective using independent frames to detect the neighboring vehicles. Next, in Part II we develop new ways for understanding the dynamics of the scene. Finally, in Part III we apply all the developed methods to accomplish higher level challenges such as segmenting moving obstacles while obtaining their rigid motion vector over the ground. More specifically, in Chapter 2 we develop a 3D vehicle detection pipeline based on a multi-branch deep-learning architecture and propose a Front (FR-V) and a Bird’s Eye view (BE-V) as 2D representations of the 3D point cloud to serve as input for training our models. Later on, in Chapter 3 we apply and further test this method on two real uses-cases, for pre-filtering moving obstacles while creating maps to better localize ourselves on subsequent days, as well as for vehicle tracking. From the dynamic perspective, in Chapter 4 we learn from the 3D point cloud a novel dynamic feature that resembles optical flow from RGB images. For that, we develop a new approach to leverage RGB optical flow as pseudo ground truth for training purposes but allowing the use of only 3D LiDAR data at inference time. Additionally, in Chapter 5 we explore the benefits of combining classification and regression learning problems to face the optical flow estimation task in a joint coarse-and-fine manner. Lastly, in Chapter 6 we gather the previous methods and demonstrate that with these independent tasks we can guide the learning of higher challenging problems such as segmentation and motion estimation of moving vehicles from our own moving perspective. Con más de 1,35 millones de muertes por accidentes de tráfico en el mundo, a principios de siglo se predijo que la conducción autónoma sería una solución viable para mejorar la seguridad en nuestras carreteras. Además la conducción autónoma está destinada a cambiar nuestros paradigmas de transporte, permitiendo reducir la congestión del tráfico, la contaminación y el coste, a la vez que aumentando la accesibilidad, la eficiencia y confiabilidad del transporte tanto de personas como de mercancías. Aunque algunos avances, como el control de crucero adaptativo, la detección de puntos ciegos o el estacionamiento automático, se han transferido gradualmente a vehículos comerciales en la forma de los Sistemas Avanzados de Asistencia a la Conducción (ADAS), la tecnología aún no ha alcanzado el suficiente grado de madurez. Se necesita una comprensión completa de la escena para que los vehículos puedan entender el entorno, detectando los elementos presentes, así como su movimiento, intenciones e interacciones. En la presente tesis doctoral, exploramos nuevos enfoques para comprender escenarios de conducción utilizando nubes de puntos en 3D capturadas con sensores LiDAR, para lo cual empleamos métodos de aprendizaje profundo. Con este fin, en la Parte I analizamos la escena desde una perspectiva estática para detectar vehículos. A continuación, en la Parte II, desarrollamos nuevas formas de entender las dinámicas del entorno. Finalmente, en la Parte III aplicamos los métodos previamente desarrollados para lograr desafíos de nivel superior, como segmentar obstáculos dinámicos a la vez que estimamos su vector de movimiento sobre el suelo. Específicamente, en el Capítulo 2 detectamos vehículos en 3D creando una arquitectura de aprendizaje profundo de dos ramas y proponemos una vista frontal (FR-V) y una vista de pájaro (BE-V) como representaciones 2D de la nube de puntos 3D que sirven como entrada para entrenar nuestros modelos. Más adelante, en el Capítulo 3 aplicamos y probamos aún más este método en dos casos de uso reales, tanto para filtrar obstáculos en movimiento previamente a la creación de mapas sobre los que poder localizarnos mejor en los días posteriores, como para el seguimiento de vehículos.", "source_url": "https:\/\/openalex.org\/W3021063730", "pdf_url": "https:\/\/upcommons.upc.edu\/bitstreams\/b8c27eee-0c12-4e12-8f18-1b7e1af52108\/download", "doi": "10.5821\/dissertation-2117-341602", "handle": "", "repository_name": "UPCommons institutional repository (Universitat Politècnica de Catalunya)", "source_record_id": "W3021063730", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 666, "title": "Human Action Recognition in Videos with Shallow and Deep Learning", "title_ar": "", "title_en": "", "author": "Thi hai hong Phan", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2019, "language": "fr", "research_domain": "computer-vision", "research_subdomain": "video-analysis", "categories": "machine-learning; deep-learning; computer-vision; cnn; video-analysis", "keywords": "", "abstract": "Reconnaissance d'actions humaines dans des vidéos avec l'apprentissage automatique Ces dernières années, la reconnaissance d’action humaine (HAR) a attiré l’attention de la recherche grâce à ses diverses applications telles que les systèmes de surveillance intelligents, l’indexation vidéo, l’analyse des activités humaines, les interactions homme-machine, et ainsi de suite. Les problèmes typiques que les chercheurs envisagent sont la complexité des mouvements humains, les variations spatio-temporelles, l'encombrement, l'occlusion et le changement des conditions d'éclairage. Cette thèse porte sur la reconnaissance automatique des actions humaines en cours dans une vidéo. Nous abordons ce problème de recherche en utilisant à la fois des approches d'apprentissage traditionnel peu profond et d'apprentissage profond.Premièrement, nous avons commencé les travaux de recherche avec des méthodes d’apprentissage traditionnelles peu profondes, fondées sur des caractéristiques créées manuellement, en introduisant un nouveau fonctionnalité appelée descripteur MOMP (Motion of Oriented Magnitudes Patterns). Nous avons ensuite intégré ce descripteur discriminant aux techniques de représentation simples mais puissantes telles que le sac de mots visuels, le vecteur de descripteurs agrégés localement (VLAD) et le vecteur de Fisher pour mieux représenter les actions. En suite l'PCA (Principal Component Analysis) et la sélection des caractéristiques (la dépendance statistique, l'information mutuelle) sont appliquées pour rechercher le meilleur sous-ensemble des caractéristiques afin d'améliorer les performances et de réduire les coûts de calcul. La méthode proposée a permis d'obtenir les résultats d'état de l'art sur plusieurs bases de données communes.Les approches d'apprentissage profond récentes nécessitent des calculs intensifs et une utilisation importante de la mémoire. Ils sont donc difficiles à utiliser et à déployer sur des systèmes aux ressources limitées. Dans la deuxième partie de cette thèse, nous présentons un nouvel algorithme efficace pour compresser les modèles de réseau de neurones convolutionnels afin de réduire à la fois le coût de calcul et l’empreinte mémoire au moment de l’exécution. Nous mesurons la redondance des paramètres en fonction de leurs relations à l'aide des critères basés sur la théorie de l'information, puis nous éliminons les moins importants. La méthode proposée réduit considérablement la taille des modèles de différents réseaux tels qu'AlexNet, ResNet jusqu'à 70% sans perte de performance pour la tâche de classification des images à grande échelle.L'approche traditionnelle avec le descripteur proposé a permis d'obtenir d'excellentes performances pour la reconnaissance de l'action humaine mais seulement sur de petits bases de données. Afin d'améliorer les performances de la reconnaissance sur les bases de données de grande échelle, dans la dernière partie de cette thèse, nous exploitons des techniques d'apprentissage profond pour classifier les actions. Nous introduisons les concepts de l'image MOMP en tant que couche d'entrée de CNN et incorporons l'image MOMP dans des réseaux de neurones profonds. Nous appliquons ensuite notre algorithme de compression réseau pour accélérer et améliorer les performances du système. La méthode proposée réduit la taille du modèle, diminue le sur-apprentissage et augmente ainsi la performance globale de CNN sur les bases de données d'action à grande échelle.Tout au long de la thèse, nous avons montré que nos algorithmes obtenaient de bonnes performances sur bases de données d'action complexes (Weizmann, KTH, UCF Sports, UCF-101 et HMDB51) avec des ressources limitées.", "source_url": "https:\/\/openalex.org\/W7148690778", "pdf_url": "", "doi": "10.70675\/fa813329z9321z4fc2za918zd4282f6ec3ac", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7148690778", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 670, "title": "Deep Learning Methods for Human Action Recognition", "title_ar": "", "title_en": "Deep Learning Methods for Human Action Recognition", "author": "Zeqi Yu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Auckland University of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "New Zealand", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "computer-vision", "research_subdomain": "video-analysis", "categories": "machine-learning; deep-learning; computer-vision; cnn; rnn; video-analysis", "keywords": "", "abstract": "Human action recognition from digital videos is a hot topic in the field of computer vision. It has a pretty assortment of applications in a myriad of fields such as video surveillance, human-computer interaction, visual information retrieval, and unmanned driving. With the exponential growth of surveillance data on the Internet in recent years, how to implement effective and efficient analysis of video data is extremely crucial. Traditional machine learning methods that only extract computable features have limitations and do not suit massive visual data, meanwhile, deep learning methods, especially convolutional neural networks, have gained great attainments in this field. The goal of human action recognition is to classify patterns so as to understand human actions from visual data and export corresponding tags. In addition to spatial correlation existing in 2D images, human actions in a video on the correlation in the temporal domain. Due to the complexity of human actions, changes of perspectives, background noises, and lighting conditions will affect the recognition. In order to solve these thorny problems, three algorithms are designed and implemented in this thesis. Based on convolutional neural networks (CNN), Two-Stream CNN, CNN+LSTM, and 3D-CNN are harnessed to identify human actions. Each algorithm is explicated and analyzed in detail. HMDB-51 dataset is employed to test these algorithms and gain the best results. Our experimental results demonstrate that the three methods have effectively identified human actions in given videos, the best algorithm thus is verified.", "source_url": "https:\/\/openalex.org\/W3157063314", "pdf_url": "https:\/\/openrepository.aut.ac.nz\/bitstreams\/ed753409-6ddd-4042-9f3a-2fff51193d63\/download", "doi": "", "handle": "", "repository_name": "Tuwhera (Auckland University of Technology)", "source_record_id": "W3157063314", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 671, "title": "Learning Interaction Grammar for Action Recognition", "title_ar": "", "title_en": "Learning Interaction Grammar for Action Recognition", "author": "Tengyu Liu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "computer-vision", "research_subdomain": "video-analysis", "categories": "deep-learning; computer-vision; video-analysis", "keywords": "", "abstract": "Video action recognition has been in the center of the stage since its introduction in 2004 [SLC04]. During the past 15 years, countless methods had been proposed to understand what human is doing in a video clip. While some works infer action label directly from pixel information, some other works propose to learn multi-level hierarchical structure that composes actions. In this work, we propose the learning of a two-layer grammar model for action recognition that is based on human object interaction. To evaluate the idea, we proposed an interaction grammar for action recognition on video. The model is evaluated by simulated annealing with MCMC and by deep learning.", "source_url": "https:\/\/openalex.org\/W2792899696", "pdf_url": "", "doi": "", "handle": "", "repository_name": "eScholarship (California Digital Library)", "source_record_id": "W2792899696", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 672, "title": "Action recognition using deep learning", "title_ar": "", "title_en": "Action recognition using deep learning", "author": "Petar Palasek", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Queen Mary University of London", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "computer-vision", "research_subdomain": "video-analysis", "categories": "deep-learning; computer-vision; cnn; video-analysis", "keywords": "", "abstract": "In this thesis we study deep learning architectures for the problem of human action recognition in image sequences, i.e. the problem of automatically recognizing what people are doing in a given video.As unlabeled video data is easily accessible these days, we first explore models that can learn meaningful representations of sequences without actually having to know what is happening in the sequences at hand.More specifically, we first explore the convolutional restricted Boltzmann machine (RBM) and show how a stack of convolutional RBMs can be used to learn and extract features from sequences in an unsupervised way.Using the classical Fisher vector pipeline", "source_url": "https:\/\/openalex.org\/W2785072843", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Queen Mary Research Online (Queen Mary University of London)", "source_record_id": "W2785072843", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 677, "title": "Transfer Learning Methods for Object and Action Recognition", "title_ar": "", "title_en": "Transfer Learning Methods for Object and Action Recognition", "author": "Samitha Herath", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Australian National University", "university_source": "repository_host", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "video-analysis", "categories": "machine-learning; computer-vision; generative-ai; ai-applications; transfer-learning; few-zero-shot; video-analysis; gan; ai-education", "keywords": "", "abstract": "Thanks to the advancement of machine learning and computer vision research we are observing remarkable improvements in object and action recognition. However, it can be seen that when there are differences (e.g., distributional, representations, semantic) in the training and test data, many learning algorithms fail to perform well or at times becomes infeasible. Nevertheless, problem setups dealing specifically with such differences are constantly being explored in literature. In general these problems are known as transfer learning problems. For instance, the objective of Domain Adaptation (DA) is to learn models that are invariant to distributional differences between train and test data (i.e., source and target domain data). Zero-Shot Learning (ZSL) is a framework for recognizing instances from classes never seen during training (i.e., missing classes in training data). In ZSL, the idea is to use related semantic information about data to workaround this data deficiency. In this thesis we present two solutions for Domain Adaptation and Zero-Shot Learning. First, we discuss our DA solution for learning an invariant Hilbert space for source and target domain data embedding. This is achieved by learning projections from individual domains to a latent space. We learn a Mahalanobis metric on this latent space to simultaneously minimizing domain disparity while maximizing a measure of discriminative power. We also show that the alignment of statistical properties (e.g., first and second order statistics) helps in improving our solution's performance, especially when there are no labeled samples from the target domain. We explore the effect of geometrical constraints such as orthogonality and positive definiteness for our model. Here, we use Riemannian optimization methods to seamlessly implement these constraints in our model training. Inspired by the success of statistical alignment, we further explore statistical alignment by min-max learning. Our motivation here is to avoid the trivial feature degeneration occurring when aligning two spaces by minimizing a statistical loss such as KL divergence. Here, we show that min-max statistical alignment is a promising solution for DA as well as ZSL. In fact, we show that min-max statistical alignment can reach comparable performance with state of the art adversarial solutions, including the ones using powerful Wasserstein GANs as the backbone. Lastly, we discuss our work on spatio-temporal transfer learning for human action recognition. In our survey on human action recognition from videos, we extensively study representation based and deep network based action recognition solutions. Here, we observe that one of the high-impact video action recognition solutions is the Two-Stream networks solution. However, the applicability of two-stream solution requires a temporal stream. In other words this restricts its applicability only to video data. In this work we explore use of transfer learning methods for devising two stream action recognition for still action images where there are no temporal information. We call this Temporal to Still Image Learning (or in short T2SIL). To explore the T2SIL framework we use existing still image and video datasets to form a new human action dataset of 40 classes. The data set contains still image actions and a video actions counterpart with the same set of classes. In this work, we explore three state of the art transfer learning solutions (i.e., Deep embedding learning, GANs and model distillation with Teacher-Student Networks) for T2SIL.", "source_url": "https:\/\/openalex.org\/W3036155538", "pdf_url": "", "doi": "10.25911\/5ee74e7c697e0", "handle": "", "repository_name": "ANU Open Research (Australian National University)", "source_record_id": "W3036155538", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 679, "title": "Robust and efficient deep learning methods for vision-based action recognition", "title_ar": "", "title_en": "Robust and efficient deep learning methods for vision-based action recognition", "author": "Yuecong Xu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Nanyang Technological University", "university_source": "repository_host", "college": "", "department": "", "country": "Singapore", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "computer-vision", "research_subdomain": "video-analysis", "categories": "machine-learning; deep-learning; computer-vision; robotics; transfer-learning; rnn; image-classification; video-analysis", "keywords": "", "abstract": "Vision-based action recognition, which performs action recognition based solely on RGB frames, has received strong research interest thanks to its wide applications in various fields, e.g. surveillance, smart homes, and autonomous driving. Significant progress has been made in vision-based action recognition thanks to the development of recognition technologies, particularly deep learning methods which have proven their effectiveness in visual recognition tasks, such as image classification. Compared to static images, videos contain additional information due to the additional temporal dimension, which includes both temporal and spatiotemporal correlation features. Therefore, the key to robust and efficient action recognition lies in the effective and efficient utilization of the temporal and spatiotemporal correlation features embedded within videos. In this thesis, we first investigate extracting temporal features in a robust and efficient manner. While methods for extracting temporal features were proposed, these methods either require the computation or estimation of optical flow, which demand high computational power and large storage resource; or extracts only linear feature along the temporal dimension which results in inferior performances. To extract temporal features without the utilization of optical flow, we propose Attentive Correlated Temporal Feature (ACTF) which leverages inter-frame correlation feature and exploits both bilinear and linear correlations between successive frames on the regional level. By excluding optical flow estimation or calculation, ACTF can be combined with any spatial feature extraction network under the two-stream structure for end-to-end training. Meanwhile, capturing long-range spatiotemporal dependencies is an effective strategy for extracting spatiotemporal correlation features. Previous works have proposed methods utilizing either hand-crafted features or stacks of convolution or recurrent modules, both of which are computationally inefficient, and cause difficulty in network optimization. While the more recent non-local block inspired by the non-local means could extract long-range dependencies without affecting the networks' optimization, it significantly increases the parameter size and computational cost of the inserted networks. To extract robust long-range dependencies more efficiently, we explore on further improving the non-local neural network by proposing a novel long-range spatiotemporal dependencies extraction module, the Pyramid Non-Local (PNL) module. It extends the original non-local block by incorporating regional feature correlations at multiple scales. PNL upscales the effectiveness of the original non-local block by additionally addressing the spatiotemporal correlations between different regions, while improving the efficiency of the original non-local block with a significant decrease in computation cost. Besides the development of recognition technologies, the progress made in vision-based action recognition could also be attributed to the development of large-scale video datasets, which enable the effective training of deep learning models. However, it could be observed that the majority of current research focuses on videos in normal illumination, partly due to the fact that current benchmark datasets for vision-based action recognition are normally collected from web videos shot mostly under normal illumination. Yet, we argue that vision-based action recognition should not be constrained in normal illuminated videos. Vision-based action recognition in dark videos are also useful in various scenarios, e.g., night surveillance and self-driving at night. Such a task has rarely been researched, partly due to the lack of sufficient datasets for such a task. To this end, this thesis bridges the gap of the lack of data and pioneers vision-based action recognition in dark videos by collecting a novel dataset: the Action Recognition in the Dark (ARID) dataset. In this thesis, the ARID dataset is analyzed thoroughly with a comprehensive benchmark of current deep learning methods. Meanwhile, though the introduction of ARID pioneers vision-based action recognition in dark videos and bridges the gap between the absence of dark video datasets with the need for research, the scale of such a dataset is relatively small compared to current large-scale video datasets. One solution for training robust models for domains with less labeled data would be to transfer models learned in well-labeled domains. However, models trained in one domain would not generalize well in the other domain due to domain shift across domains, which is presented as distribution discrepancy between different domains. Domain adaptation (DA) approaches address domain shifts and enable networks to be applied to different scenarios.", "source_url": "https:\/\/openalex.org\/W3214258728", "pdf_url": "", "doi": "10.32657\/10356\/153169", "handle": "", "repository_name": "DR-NTU (Nanyang Technological University)", "source_record_id": "W3214258728", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 680, "title": "Leveraging deep learning for visual understanding of videos", "title_ar": "", "title_en": "Leveraging deep learning for visual understanding of videos", "author": "Clement Xian Ren Tan", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Nanyang Technological University", "university_source": "repository_host", "college": "", "department": "", "country": "Singapore", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; robotics; self-semi-supervised; predictive-analytics; transformers; multimodal; image-classification; object-detection; image-segmentation; video-analysis; llm", "keywords": "", "abstract": "In the field of Computer Vision (CV), the pursuit of human-level recognition and reasoning of visual scenes has been a long-standing aspiration. Over the past decade, significant contributions to the progress of CV have been made by deep learning, facilitated by the availability of big data and increased computational power. In CV, visual understanding encompasses not only the fundamental aspect of recognition, which involves identifying and categorizing objects and patterns within visual data, but also reasoning, which involves higher-level cognitive processes such as inferring relationships, predicting outcomes, and drawing meaningful conclusions based on the observed visual information. Human-like recognition capabilities across a variety of visual recognition tasks such as image classification, object detection, semantic segmentation, and instance segmentation, have been achieved by machines. The huge success of deep learning for visual recognition tasks has prompted researchers to tackle visual reasoning tasks, which are more challenging. The recognition and reasoning of objects within a scene is of paramount importance to applications such as human-robot collaborations and autonomous vehicles. This thesis aims at addressing the issues associated with current deep learning models for visual understanding and will focus on three tasks: (1) video object segmentation, (2) abductive action inference, and (3) action-conditioned scene graph prediction. Firstly, video object segmentation is an object tracking task that traditionally relies on supervised learning and necessitates extensive annotated datasets for deep learning model training. It is confronted with issues of repetitiveness and impracticality due to the manual labeling of these vast datasets. In addition, tracking objects within videos presents a unique challenge caused by appearance changes and occlusion across video frames. The work proposed in this thesis explores the potential of self-supervised learning for video object segmentation, aiming to harness freely available internet data such as YouTube videos for model training. Different from existing self-supervised learning approaches for video object segmentation that model pixel-to-pixel correspondence, this research shifts the focus towards modeling superpixel-to-superpixel correspondence. To achieve this, a novel approach which involves the tracking of superpixels between video frames through an attention mechanism trained within an end-to-end self-supervised framework is proposed. This approach aims to enhance the performance of existing self-supervised video object segmentation models by leveraging deep learning techniques and mitigating the limitations associated with supervised learning using large-scale annotated datasets. It also aims to enhance existing self-supervised models with the benefits of superpixels. Secondly, this thesis introduces the complex task of abductive action inference to assess the abductive reasoning abilities of deep learning models in comprehending visual scenes. A set of innovative object-relational models designed specifically for this task is presented and compared with existing state-of-the-art image, video, and vision-language models. These models are tasked with abducting the human-performed action that led to the scene depicted in an image or snapshot. The experiments demonstrate the potential of deep learning models in performing abductive action inference. This research is vital in advancing our understanding of the reasoning capacities of deep learning models, representing a significant stride toward attaining human-level reasoning capabilities in visual comprehension. Lastly, in action-conditioned scene graph prediction, the task involves predicting scene graph relations for a future state based on an initial state and an action. Instead of predicting visual representations for future states, each state is organized into a scene graph comprising human-object relationship triplets, thereby succinctly encapsulating the dynamics within the scene. Notably, this task remains underdeveloped in visual understanding research, with prior video scene graph generation approaches largely overlooking actions as a crucial signal. Actions applied to a preconditioned state generate an effect state embodying their anticipated outcomes. Human cognitive reasoning allows the inference of action consequences based on the initial scene context. To address this challenge, this thesis introduces the Action-conditioned Scene Graph dataset. Furthermore, this thesis proposes the AERT (Action-conditioned Effect Relational Transformer) model, designed to capture scene relations and action context to effectively predict future scene graph relations.", "source_url": "https:\/\/openalex.org\/W4402201451", "pdf_url": "", "doi": "10.32657\/10356\/179521", "handle": "", "repository_name": "DR-NTU (Nanyang Technological University)", "source_record_id": "W4402201451", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 686, "title": "Trustworthiness in Vertical Federated Learning: Privacy attacks, Robustness Vulnerabilities, and Secure Protocols", "title_ar": "", "title_en": "Trustworthiness in Vertical Federated Learning: Privacy attacks, Robustness Vulnerabilities, and Secure Protocols", "author": "Duanyi Yao", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "federated-learning", "categories": "machine-learning; medical-ai; responsible-ai; ai-applications; federated-learning; fairness; ai-cybersecurity", "keywords": "", "abstract": "Vertical Federated Learning (VFL) enables organizations with distinct features of the same data samples to collaboratively train machine learning models without directly sharing raw data. While VFL offers strong potential for privacy-preserving analytics in domains such as finance and healthcare, its trustworthiness remains insufficiently understood. This thesis provides a systematic study of VFL trustworthiness along three dimensions: privacy, robustness, and secure protocol design. First, we investigate privacy vulnerabilities by introducing URVFL, an undetectable data reconstruction attack. By generating malicious gradients through a discriminator with auxiliary classifier, URVFL achieves state-of-the-art reconstruction fidelity while evading existing detection mechanisms, revealing the fragility of current defenses. Second, we analyze robustness vulnerabilities by examining adversarial example attacks in VFL inference. We formulate adaptive corruption of client-server channels as an online optimization problem and propose the E-TS algorithm, which leverages empirical Thompson sampling to identify optimal corruption patterns with provable regret bounds. Our results demonstrate that adaptive adversaries can severely compromise VFL robustness. Finally, we design FedVS, a synchronous split VFL protocol that simultaneously provides straggler resilience and information-theoretic privacy. Through coding-based secret sharing, FedVS losslessly aggregates embeddings even with stragglers while preventing leakage to curious servers and colluding clients. This thesis demonstrates that trustworthy VFL requires a holistic approach addressing privacy, robustness, and protocol-level vulnerabilities simultaneously. The proposed attacks expose critical weaknesses that must be addressed before large-scale deployment, while our secure protocol provides a constructive foundation for privacy-preserving collaborative learning. These contributions advance understanding of trustworthiness in distributed machine learning and establish principled design criteria for secure VFL systems.", "source_url": "https:\/\/openalex.org\/W7124459682", "pdf_url": "", "doi": "10.14711\/thesis-hdl167715", "handle": "", "repository_name": "", "source_record_id": "W7124459682", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 695, "title": "Designing for Privacy in Tabular Federated Learning", "title_ar": "", "title_en": "Designing for Privacy in Tabular Federated Learning", "author": "Ivo Osterberg Nilsson ; Maximilian Birr Engvall", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Jönköping University", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; responsible-ai; ai-applications; predictive-analytics; transformers; federated-learning; fairness; ai-cybersecurity", "keywords": "", "abstract": "In federated learning (FL), private data leakage through gradient inversion depends strongly on design choices and data properties. We study this risk under an honest-but-curious server threat model, with a focus on clinical tabular data from MIMIC-IV and complementary benchmark tasks for binary classification, multiclass classification, and regression. Unlike image reconstruction, tabular reconstruction cannot be judged well by visual inspection. Evaluation must account for numerical tolerances, categorical exact matches, baseline recoverability, and the difference between feature level recovery and exact row recovery. We evaluate FedSGD gradients and FedAvg model deltas using an exposure aligned training protocol, where attacked models are compared after matched client data exposure rather than matched communication rounds. We compare MLP, ResNet, and FT-Transformer models, and isolate architecture effects further through an MLP grid over width, depth, activation, normalization, and dropout. The results show that local aggregation and client data exposure are among the strongest determinants of tabular gradient leakage. Small client batches and updates representing few distinct records are most vulnerable, while stronger local aggregation and greater client data exposure usually reduce reconstruction in ways that vary across architectures and datasets. FT-Transformer is consistently harder to invert than the MLP and ResNet baselines, suggesting that embedded categorical representations change the geometry of tabular gradient inversion. At the same time, the MLP architecture grid shows that reconstructability also changes substantially within a single model family, so architecture affects privacy beyond the transformer comparison alone. Aggregate reconstruction accuracy can overstate record recovery, especially on MIMIC-IV, where sparse and repeated clinical features produce a strong client marginal prior. Strict exact match results show that complete row recovery is concentrated in the most exposed one-hot-encoded settings, while FT-Transformer can have nonzero feature recovery without exact row recovery. Overall, the findings show that gradient inversion risk in tabular FL is shaped by model architecture, FL training configuration, data exposure, and input data properties.", "source_url": "https:\/\/openalex.org\/W7165261472", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Jonkoping University Library (Jönköping University)", "source_record_id": "W7165261472", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 705, "title": "Privacy and Security Enhanced Federated Learning Framework Design", "title_ar": "", "title_en": "Privacy and Security Enhanced Federated Learning Framework Design", "author": "Yasaman Pakdel", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "medical-ai", "research_subdomain": "federated-learning", "categories": "machine-learning; medical-ai; responsible-ai; federated-learning; fairness", "keywords": "", "abstract": "Federated learning, while promising for its ability to train models across decentralized data sources, presents both security challenges and solutions.Its security depends on how well these challenges are addressed.On one hand, federated learning can be secure when proper privacy-preserving techniques are employed, such as encryption, secure aggregation, and diferential privacy, to protect data and model updates.Participant authentication, secure communication, and compliance with data protection regulations further contribute to its security.However, federated learning may also be vulnerable if not carefully implemented, potentially exposing sensitive data or risking model integrity.By encrypting and processing data using Crypten, federated learning participants can collaborate on model training without exposing their raw data, ensuring data privacy and confdentiality.Crypten's features, such as secure aggregation and encrypted computation, further bolster the trustworthiness of federated learning, making it an indispensable tool for applications in healthcare, fnance, and other sectors where sensitive data protection is paramount.In an era of growing data privacy concerns, Crypten empowers federated learning to thrive in a secure and privacy-preserving manner.In this project, we embarked on a comprehensive exploration of machine learning methodologies, starting with traditional plaintext machine learning techniques.Subsequently, we", "source_url": "https:\/\/openalex.org\/W4400228779", "pdf_url": "http:\/\/scholarworks.calstate.edu\/downloads\/70795g815", "doi": "10.46569\/9s161f47k", "handle": "", "repository_name": "", "source_record_id": "W4400228779", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 712, "title": "Deep Learning-Based Medical Image Analysis for Enhanced Diagnosis and Severity Assessment of Viral Pneumonia", "title_ar": "", "title_en": "Deep Learning-Based Medical Image Analysis for Enhanced Diagnosis and Severity Assessment of Viral Pneumonia", "author": "Enshaei, Nastaran", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Concordia University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; predictive-analytics; image-segmentation; medical-imaging", "keywords": "", "abstract": "Viral pneumonia remains a critical global health concern due to its potential for rapid transmission and widespread outbreaks, as highlighted by the COVID-19 pandemic. Timely and accurate diagnosis, severity assessment, and prognosis are essential for effective patient care and public health management. Chest imaging modalities such as chest X-rays (CXRs) and computed tomography (CT) scans play a central role in diagnosing and managing viral pneumonia. However, interpreting these images is often resource-intensive and subject to inter-observer variability, especially during surges in clinical demand. Deep learning (DL) holds promise for automating image analysis, detecting subtle radiologic features, and enhancing clinical decision-making. Nevertheless, real-world deployment faces persistent challenges, including limited access to large-scale well-annotated datasets, domain shifts, and the heterogeneous presentation of disease. To address these challenges, this thesis advances DL-based medical image analysis for the diagnosis and severity assessment of viral pneumonia, with a specific focus on COVID-19. Key contributions include: (i) Employing game-theoretic data valuation (Data Shapley) to identify mislabeled samples in chest image training datasets; (ii) Introducing a domain-adaptive strategy guided by model prediction confidence to improve the generalization of DL segmentation models; (iii) Developing robust DL frameworks for precise segmentation of COVID-19 lesions in chest CT scans, accounting for their diverse visual characteristics and spatial variability; and (iv) Designing a DL-based diagnostic and severity scoring system for early-stage COVID-19 using baseline CXRs, with performance comparable to expert radiologists. Together, these contributions advance DL-assisted tools for the diagnosis and severity assessment of viral pneumonia in clinical practice.", "source_url": "https:\/\/openalex.org\/W7112797530", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Spectrum Research Repository (Concordia University)", "source_record_id": "W7112797530", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 720, "title": "Machine Learning Algorithms for Computer-aided Diagnosis of Myocardial Infarction from 2D Echocardiography and COVID-19 from Chest X-rays", "title_ar": "", "title_en": "Machine Learning Algorithms for Computer-aided Diagnosis of Myocardial Infarction from 2D Echocardiography and COVID-19 from Chest X-rays", "author": "Aysen Degerli Ahishali", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Tampere University", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; ai-applications; image-segmentation; medical-imaging; ai-cybersecurity", "keywords": "", "abstract": "The importance of e-health has drastically increased as the world has engaged sharply in the digital world, which provoked computer-aided diagnosis (CAD) algorithms to become a necessity for accurate and automatic detection and differential diagnosis of various diseases and medical conditions. Recent advancements in machine learning brought novel solutions for more complex clinical tasks that improved medical diagnostics. Accordingly, machine and deep learning methods are widely utilized in CAD-based techniques, which analyze medical data by performing pattern recognition to reveal any anomalies to guide medical doctors and experts in clinical tasks. However, in the development phase of CAD algorithms, there are several limitations and challenges regarding the size, quality, and availability of medical data for analysis using machine learning. Hence, there is an urgent need for accurate, reliable, and robust CAD algorithms to overcome the limitations and challenges in medical diagnostics. This thesis focuses on providing machine learning algorithms for CAD-based detection and assessment of myocardial infarction (MI) and coronavirus disease 2019 (COVID-19) using apical 4-chamber (A4C) and apical 2-chamber (A2C) view 2D echocardiography recordings and chest X-ray (CXR) images, respectively. The hypothesis defined in this thesis is that the proposed CAD algorithms using machine learning can provide assistive and visualization tools in clinical practice for the detection and assessment of MI and COVID-19. Accordingly, the thesis contributions are categorized into computer-aided diagnosis of MI and COVID-19, where task-specific solutions are developed for the assessments. The most severe manifestation of coronary artery disease is myocardial infarction, which is colloquially known as heart attack that leads to irreversible necrosis of the myocardium and even death. The detection of MI requires the assessment of regional wall motion abnormality of the left ventricle that can be primarily monitored and evaluated with 2D echocardiography. This thesis contributes to a pioneer 2D echocardiography dataset, HMC-QU for the detection of MI over single-view and multi-view echocardiography. Accordingly, another contribution of this thesis is a generic pseudo-labeling annotation approach that accelerates and enhances pixel-level manual annotations by preventing the subjective and time-consuming annotation process. In single-view and multi-view echocardiography assessments, this thesis contributes to a feature engineering phase, where myocardial segment features are extracted and used to train several traditional machine learning models and one-class classification methods reaching the highest sensitivity levels of 88.75% in single-view and 90.91% in multi-view echocardiography. The assessment of pulmonary regions plays a vital role in COVID-19 clinical diagnosis and prevention of contamination to the public. This thesis contributes to the composition of the pioneer and the largest CXR dataset, called QaTa-COV19, for the recognition, early detection, and pneumonia segmentation of COVID-19. Accordingly, another contribution of this thesis is the human-machine collaboration approach for the annotation of COVID-19 pneumonia regions for the segmentation. The thesis contributes to several machine learning models and solutions to perform a reliable detection and localization of COVID-19 pneumonia in the lungs using segmentation neural networks, direct detection from a proposed operational neural network, and representation-based classification models. The findings of this thesis are promising in computer-aided diagnosis of MI and COVID-19, where the proposed solutions reached high sensitivity levels of 90.9% and 98.6%, respectively preventing subjective and expert-dependent assessments and detection in 2D echocardiography and CXR imaging. The last contribution of this thesis is the developed visualization and assistive tools, where color-coded trackings on echocardiography recordings, myocardial infarction displacement curves, COVID-19 infection maps, and machine learning-based detection and annotation tools serve medical doctors in medical diagnostics addressing visual enhancements over 2D echocardiography and CXR imaging.", "source_url": "https:\/\/openalex.org\/W7133005963", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Tampere University Institutional Repository (Tampere University)", "source_record_id": "W7133005963", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 725, "title": "Classification of chest diseases from x-ray images using deep learning", "title_ar": "", "title_en": "Classification of chest diseases from x-ray images using deep learning", "author": "Arifa Khatun", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Eastern Finland", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; predictive-analytics; medical-imaging", "keywords": "", "abstract": "In today's modern age, the field of Artificial Intelligence and Health Care are inseparable.Every now and then, doctors and patients use various technology to diagnose diseases.But still, interpreting or identifying diseases through different medical tests such as X-ray or MRI cannot be fully done using artificial intelligence due to the lack of good output from various models.This paper tried to tackle this problem of classifying three diseases that a professional can identify through a chest X-ray image.Prior to this work, many had done similar works but only a few were able to get prediction accuracy over 90%.So, this paper explored that area to get better prediction result and by fitting ResNet50, DenseNet121, MobileNetV2, InceptionNetV3 models, it was seen that ResNet50 outperformed other three models with over 98% accuracy.The result shows that deep learning models can do well classifying diseases from a test image with a very little margin of error.", "source_url": "https:\/\/openalex.org\/W7215672098", "pdf_url": "https:\/\/erepo.uef.fi\/bitstreams\/ff3e94fd-ba08-4c15-94d8-a48c0cc1e702\/download", "doi": "", "handle": "", "repository_name": "UEF eRepo (University of Eastern Finland)", "source_record_id": "W7215672098", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 735, "title": "Adaptive Deep Learning for Vision and Multimodal Understanding with Heterogeneous Data", "title_ar": "", "title_en": "Adaptive Deep Learning for Vision and Multimodal Understanding with Heterogeneous Data", "author": "Youwei Liang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; medical-ai; transfer-learning; few-zero-shot; predictive-analytics; transformers; multimodal; image-segmentation; llm", "keywords": "", "abstract": "Large-scale pretraining has produced powerful deep learning representations, but real-world data rarely match the clean, uniform assumptions of standard transfer learning. This dissertation uses heterogeneous data to describe settings in which information differs in modality, representation, relevance, quality, or frequency across inputs, outputs, tasks, or supervision. Examples include molecules represented as graphs, images, strings, and text; images containing both informative and redundant regions; noisy or incomplete labels; and rare but important classes. Applying pretrained models in these settings requires deciding which information to process, how to connect different forms of information, and how much to trust each supervision signal. This dissertation develops adaptive learning as a unifying approach through three paradigms: adaptive selection focuses computation on useful input components; adaptive alignment maps heterogeneous modalities, tasks, and outputs into a shared predictive space; and adaptive supervision adjusts imperfect or imbalanced training signals. The first contribution, EViT, demonstrates adaptive selection for efficient Vision Transformers. Standard Vision Transformers devote equal computation to all image patches despite substantial redundancy. EViT uses class-token attention to identify important tokens, preserves them in subsequent layers, and fuses inattentive tokens into a compact representation rather than discarding their context. This progressive token reorganization reduces computation while maintaining accurate visual recognition. The second contribution, DrugChat, demonstrates adaptive alignment for multimodal molecular prediction. Molecules can be represented as chemical graphs, rendered images, SMILES strings, and natural-language prompts, while predictions may require labels, probabilities, categorical decisions, or free-form descriptions. DrugChat encodes molecular graphs and images with pretrained encoders, projects them into the latent space of an instruction-following language model, and conditions generation on molecular structure and task prompts. It thereby aligns heterogeneous inputs in a shared representation space and expresses diverse prediction targets through a common language-generation interface. The third contribution, RobPicker, demonstrates adaptive supervision for cryo-electron tomography particle picking, where annotations are expensive, incomplete, noisy, and highly imbalanced across macromolecular species. Its meta-learning framework jointly trains a segmentation network, a label correction network, and a data reweighting network. Guided by cleaner validation data, these components refine imperfect labels and adapt each subtomogram's contribution to the loss, giving greater influence to rare or small particle classes. Together, EViT, DrugChat, and RobPicker apply adaptation at complementary levels of the learning pipeline: selecting what data to process, aligning heterogeneous representations and outputs, and learning from imperfect supervision.", "source_url": "https:\/\/openalex.org\/W7222391945", "pdf_url": "https:\/\/escholarship.org\/content\/qt3h17286t\/qt3h17286t.pdf", "doi": "", "handle": "", "repository_name": "eScholarship (California Digital Library)", "source_record_id": "W7222391945", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 737, "title": "Advancing Computational Pathology with Multimodal Deep Learning", "title_ar": "", "title_en": "Advancing Computational Pathology with Multimodal Deep Learning", "author": "Md. Mamunur Rahaman", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; medical-ai; responsible-ai; few-zero-shot; self-semi-supervised; predictive-analytics; multimodal; llm; medical-imaging; xai", "keywords": "", "abstract": "Histopathological assessment of tissue biopsies remains the cornerstone of cancer diagnosis, yet it is limited by diagnostic subjectivity and an incomplete view of the molecular heterogeneity that drives outcomes. This dissertation presents a suite of artificial intelligence (AI) frameworks that integrate deep learning, computer vision, and multimodal data to enhance diagnostic precision, prognostication, and therapeutic guidance across multiple cancer settings. The research first establishes a method for improving classification robustness through HistopathAI, a framework employing supervised contrastive learning and a hybrid deep feature fusion architecture. This approach yields more discriminative and generalizable feature representations, particularly in the context of class imbalance common in medical datasets. Building on this, the work demonstrates the power of integrating visually interpretable features from the tumor microenvironment. An AI-driven biomarker for Head and Neck Squamous Cell Carcinoma (HNSCC), which quantifies the spatial architecture of immune and stromal cells, was developed, and its prognostic value was validated on independent patient cohorts, confirming its strong potential for clinical risk stratification. A central contribution is the development of “virtual spatial transcriptomics” that bridges histology and genomics. The BrSTNet framework predicts spatial gene expression from routine H&E images in breast cancer, establishing a strong benchmark for image-to-transcriptome inference. This capability is translated to a direct clinical application in prostate cancer with ST-DoxPCa, a predictive signature for docetaxel benefit. The validation strategy was explicitly tailored to disease context: for the RTOG0521 trial (localized disease), a prognostic model trained independently in the Cleveland Clinic prostatectomy cohort (CCF) was applied in a fully blinded manner, without refitting to stratify patients and assess treatment effect; for the CHAARTED trial (metastatic disease), a separate treatment-response model was trained exclusively on the standard-of-care (ADT-only) arm. In both settings, ST-DoxPCa stratified patients into subgroups that either derived a significant survival benefit from docetaxel or showed no advantage, supporting treatment de-escalation for likely non-responders. Finally, to address the critical challenge of data scarcity, the thesis introduces MR-PHE, a novel framework for zero-shot learning in histopathology. By leveraging vision-language models with a multi-resolution, prompt-guided approach, MR-PHE enables accurate classification without task-specific training data, often outperforming fully supervised methods and offering a scalable solution for rare diseases. Collectively, these contributions advance the frontier of computational pathology, delivering a suite of validated, AI-driven tools that translate complex tissue-based data into clinically actionable insights. The methodologies developed herein represent a significant step toward a new paradigm of precision oncology, where patient care is personalized based on a deeper, quantitative understanding of tumor biology derived from standard pathology slides.", "source_url": "https:\/\/openalex.org\/W7116859814", "pdf_url": "", "doi": "10.26190\/unsworks\/31906", "handle": "", "repository_name": "UNSWorks (University of New South Wales, Sydney, Australia)", "source_record_id": "W7116859814", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 739, "title": "Multimodal AI for 3D-language understanding", "title_ar": "", "title_en": "Multimodal AI for 3D-language understanding", "author": "Andrea Amaduzzi", "supervisor": "", "degree_type": "phd", "degree_evidence": "اسم المستودع يخص أطاريح الدكتوراه: AMS Dottorato Institutional Doctoral Theses Repository (University of Bologna)", "university": "University of Bologna", "university_source": "repository_host", "college": "", "department": "", "country": "Italy", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "deep-learning; computer-vision; nlp; multimodal; llm", "keywords": "", "abstract": "Understanding the world through natural language is a fundamental challenge in computer vision, requiring models to bridge the gap between geometric representations and semantic understanding. While recent advances in deep learning have enabled impressive progress in multimodal tasks, effectively connecting 3D data with language remains difficult due to the computational complexity of 3D representations and the scarcity of high-quality paired 3D-language datasets. In this Thesis, we focus on developing novel deep learning techniques that can efficiently process and describe 3D data through natural language. First, we address the critical lack of high quality 3D-language datasets and reliable evaluation methods for text-to-3D generation by introducing a benchmark comprising GPT2Shape, a dataset of 3D models with high-quality language annotations, CrossCoherence, a novel quantitative metric for evaluating text-to-3D coherence, and HST, a human-validated test set. This benchmark demonstrates superior performance compared to existing 2D-based evaluation metrics. In the second part of this Thesis, we introduce NeRF-language models — a new paradigm of Multimodal Large Language Models that directly process the weights of Neural Radiance Fields (NeRFs) rather than explicit representations like images or point clouds materialized from them. We show that this approach enables holistic 3D understanding without computationally expensive rendering, achieving state-of-the-art performance on the first NeRF-language datasets. Finally, we propose a method for computing spatially-aware tokens directly from the weights of a NeRF, enabling fine-grained spatial understanding and detailed scene comprehension. Our resulting model demonstrates strong capabilities in spatial reasoning on multi-object scenes, with promising generalization to real-world scenarios.", "source_url": "https:\/\/openalex.org\/W7172489949", "pdf_url": "", "doi": "10.48676\/unibo\/amsdottorato\/12771", "handle": "", "repository_name": "AMS Dottorato Institutional Doctoral Theses Repository (University of Bologna)", "source_record_id": "W7172489949", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 740, "title": "Deep learning and natural language processing for computer aided diagnosis", "title_ar": "", "title_en": "Deep learning and natural language processing for computer aided diagnosis", "author": "Sadam Hussain", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Tecnológico de Monterrey", "university_source": "repository_host", "college": "", "department": "", "country": "Mexico", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; medical-ai; responsible-ai; predictive-analytics; transformers; multimodal; llm; medical-imaging; xai", "keywords": "", "abstract": "Multimodal artificial intelligence (AI) is a cutting-edge technique that integrates diverse modalities, such as imaging and textual data, to enhance classification and regression tasks. This dissertation focuses on the integration, comparison, and evaluation of multimodal AI for breast cancer diagnosis and prognosis. To achieve these objectives, we curated a comprehensive multimodal dataset comprising digital mammograms and corresponding radiological reports. Leveraging this dataset, we introduced and assessed various state-of-the-art (SOTA) multimodal techniques for three key tasks: breast cancer classification, reduction of false-positive biopsies with explainable AI (XAI), and short- term (5-year) risk prediction of breast cancer.In this work, we also introduced a benchmark dataset of radiological reports from breast cancer patients and provided baseline performance evaluations using SOTA machine learning (ML), deep learning (DL), and large language models (LLMs) for BI-RADS category classification. Our approach evaluated the performance of diverse SOTA multimodal architectures, including ResNet, VGG, E!cientNet, MobileNet, and Vision Transformers (ViT). For textual data processing, we employed both general-purpose and domain-specific pretrained LLMs such as BERT, bioGPT, ClinicalBERT, and DeBERTa, which were also integrated into multimodal architectures for enhanced lassification.Notably, our proposed multiview multimodal feature fusion (MMFF) architecture, combining SE-ResNet50 with an artificial neural network (ANN), achieved an AUC of 0.965 for breast cancer classification, significantly outperforming both single-modal and multimodal SOTA architectures. For reducing unnecessary breast biopsies, our multimodal approach achieved an AUC of 0.72, showcasing its clinical utility in minimizing patient burden. Moreover, our ViT and bioGPT-based multimodal architecture achieved an AUC of 0.77 for short-term risk prediction, outperforming the SOTA MIRAI model, which achieved an AUC of 0.59 on our in-house dataset. This work highlights the potential of multimodal AI in advancing breast cancer diagnosis and prognosis, demonstrating its superiority over traditional and unimodal approaches across multiple critical tasks.", "source_url": "https:\/\/openalex.org\/W7225556007", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Repositorio del Tecnológico de Monterrey (Tecnológico de Monterrey)", "source_record_id": "W7225556007", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 741, "title": "Exploring Multimodal Deep Learning Approaches for Coronary Artery Disease Diagnosis", "title_ar": "", "title_en": "Exploring Multimodal Deep Learning Approaches for Coronary Artery Disease Diagnosis", "author": "Popov, Maxim", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United Arab Emirates", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; medical-ai; self-semi-supervised; multimodal; llm; medical-imaging", "keywords": "", "abstract": "Coronary Artery Disease (CAD) is a prevalent cause of death in the modern world, with about 20 million victims every year. Limited annotated data, inter-observer variability, and the intrinsic complexity of angiographic imaging challenge its diagnosis. This thesis presents a novel deeplearning framework to enhance coronary artery disease (CAD) diagnosis through multimodal analysis of Xray coronary angiograms. We develop several interrelated contributions to address these issues that combine advanced contrastive learning and vision-language modeling.", "source_url": "https:\/\/openalex.org\/W7106339970", "pdf_url": "", "doi": "10.82419\/166", "handle": "", "repository_name": "MBZUAI iRep", "source_record_id": "W7106339970", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 745, "title": "Learning semantic representations of language, vision and code for visualization-oriented dialogue", "title_ar": "", "title_en": "Learning semantic representations of language, vision and code for visualization-oriented dialogue", "author": "Henrik Voigt", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; generative-ai; responsible-ai; expert-systems; transfer-learning; self-semi-supervised; transformers; multimodal; llm; fairness", "keywords": "", "abstract": "Data visualization is essential for navigating the vast volume of data that defines our time, letting humans use perception to make sense of complex datasets. Yet creating visualizations requires mastery of specialized programming languages, producing a gap between the analytical intention in the user's mind and its implementation in executable code. This raises the cognitive load on experts, who cannot concentrate fully on visual exploration while also writing the visualization program, and it creates a barrier for non-experts, hindering widespread access to insights from data. Bridging this gap is a central challenge for Visualization-oriented Natural Language Interfaces (V-NLIs), which translate human intent into executable code. This work decomposes the problem into two interrelated challenges: understanding, which builds rich, contextual representations of user intent, and generation, which reliably synthesizes visualization programs under strict real-time constraints. The thesis addresses both through five interconnected contributions. It begins with a survey of 119 papers, revealing that while the community focuses on sophisticated interactive exploration, the underlying technology relies on classical, rule-based methods, motivating a turn to deep learning. KeywordScape, a visual analytics system, makes the contextualized representations of transformer models visible and interactive, showing they can themselves resolve ambiguity and enable fine-grained exploration. Paparazzi extends this to multimodal representations, revealing that large vision-language models such as CLIP exhibit surprising deficiencies in 3D viewpoint understanding, and introduces a data-efficient contrastive fine-tuning method to improve geometric grounding. Finally, VIST5, a practical V-NLI built on a retrieval-augmented architecture around a small language model, shows that complex visualization code can be synthesized in real time while ensuring data privacy and modular extensibility.", "source_url": "https:\/\/openalex.org\/W7170734941", "pdf_url": "", "doi": "10.22032\/dbt.71264", "handle": "", "repository_name": "Common Library Network (Der Gemeinsame Bibliotheksverbund)", "source_record_id": "W7170734941", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 747, "title": "Towards Natural Language Understanding using Multimodal Deep Learning", "title_ar": "", "title_en": "Towards Natural Language Understanding using Multimodal Deep Learning", "author": "Siegfried Bös", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Delft University of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Netherlands", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "deep-learning; computer-vision; nlp; ai-applications; multimodal; object-detection; video-analysis; speech-recognition; ai-agriculture", "keywords": "", "abstract": "This thesis describes how multimodal sensor data from a 3D sensor and microphone array can be processed with deep neural networks such that its fusion, the trained neural network, is a) more robust to noise, b) outperforms unimodal recognition and c) enhances unimodal recognition in absence of multimodal data. We built a framework for a complete workflow to experiment with multimodal sensor data ranging from recording (with Kinect 3D sensor), labeling, 3D signal processing, analysing and replaying. We also built three custom recognizers (automatic speech recognizer, 3D object recognizer and 3D gesture recognizer) to convert the raw sensor streams to decisions and feed this to the neural network using a late fusion strategy. We recorded 25 particpants performing 27 unique verbal and gestural interactions (intents) with objects and trained the neural network using a supervised strategy. We proved that the framework works by building a deep neural networks assisted speech recognizer that performs approximately 5% better with multimodal data at 20 dB SnR up to 61% better with multimodal data at -5 dB SnR while performing identical to the individual recognizer when fed a unimodal datastream. Analysis shows that performance gain in low acoustic noise is due to true fusion of classifer results while gain at high acoustic noise is due to absence of speech results as it cannot detect speech events anymore, while the gesture recognizer is not affected. The impact of this thesis is significant for computational linguists and computer vision researchers as it describes how practical issues with (real and) real-time data can be solved such as dealing with sensor noise, GPU offloading for computational performance, 3D object and hand tracking. The speech-, object- and gesture recognizers are not state-of-the-art and the small vocabulary with 27 unique phrases and 9 objects can be considered a preliminary experiment. The main contributions of this thesis project are a) validated multimodal fusion framework and workflow for embodied natural language understanding named MASU, b) 600GB, 2,5 hour labelled multimodal database with synchronous multi channel audio and 3D video, c) algorithm for 3D hand-object detection and tracking, d) recipe to train a deep neural network model for multimodal fusion and e) demontrate MASU in practical real-time scenario.", "source_url": "https:\/\/openalex.org\/W2758142123", "pdf_url": "https:\/\/repository.tudelft.nl\/file\/File_91e1bf2c-74c9-43b0-bba6-3902d2b876f9", "doi": "", "handle": "", "repository_name": "Research Repository (Delft University of Technology)", "source_record_id": "W2758142123", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 751, "title": "Multimodal Deep Learning for Zero-Shot Action Recognition", "title_ar": "", "title_en": "Multimodal Deep Learning for Zero-Shot Action Recognition", "author": "Nan WU", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Japan", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; few-zero-shot; cnn; multimodal; video-analysis; llm", "keywords": "", "abstract": "Advancements in machine learning have improved the performance of machine learning (ML) models.However, larger training datasets present challenges, such as prolonged labeling and training times.One potential solution is zero-shot learning, which enables models to identify unfamiliar entities without extensive training data.This not only saves time and effort but also allows easy integration of new categories.However, its efficacy requires enhancement for broad real-world applications.This paper presents three methods to optimize the accuracy of zero-shot action recognition (ZSAR).1. Three-stream graph convolutional networks (TS-GCN):Current models often handle only one data type, limiting their information leverage.Our proposed two-stream network processes both RGB image data and optical flow data, which captures object movement.Each branch calculates video classification separately, and a weighted average determines the final classification.2. Improving ZSAR using human instruction with text description: The video-text matching model (TVMM) model gauges the matching degrees between text features and video content.We annotated text features for each action, and the TVMM model used these for action recognition.The results showed improved action recognition accuracy.3. Zero-shot action recognition with ChatGPT-based instruction: Manual text feature annotation improved accuracy but at a labor cost.To alleviate this, we used ChatGPT for automated text feature generation.This eliminated the need for human annotation.While achieving high recognition accuracy with reduced labor costs, it was slightly lower than previous methods but still suitable for various real-life applications.We propose three innovative methods that optimize zero-shot action recognition.The experiments demonstrate the effectiveness of our approach.", "source_url": "https:\/\/openalex.org\/W7207547986", "pdf_url": "https:\/\/opac.ll.chiba-u.jp\/da\/curator\/900122990\/WLA_0144.pdf", "doi": "10.20776\/900122990", "handle": "", "repository_name": "Institutional Repositories DataBase (IRDB)", "source_record_id": "W7207547986", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 757, "title": "Leveraging Multimodal Perspectives to Learn Common Sense for Vision and Language Tasks", "title_ar": "", "title_en": "Leveraging Multimodal Perspectives to Learn Common Sense for Vision and Language Tasks", "author": "Lin Xiao", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "nlp", "research_subdomain": "xai", "categories": "machine-learning; deep-learning; computer-vision; nlp; responsible-ai; predictive-analytics; multimodal; xai", "keywords": "", "abstract": "Learning and reasoning with common sense is a challenging problem in Artificial Intelligence (AI). Humans have the remarkable ability to interpret images and text from different perspectives in multiple modalities, and to use large amounts of commonsense knowledge while performing visual or textual tasks. Inspired by that ability, we approach commonsense learning as leveraging perspectives from multiple modalities for images and text in the context of vision and language tasks. Given a target task (e.g., textual reasoning, matching images with captions), our system first represents input images and text in multiple modalities (e.g., vision, text, abstract scenes and facts). Those modalities provide different perspectives to interpret the input images and text. And then based on those perspectives, the system performs reasoning to make a joint prediction for the target task. Surprisingly, we show that interpreting textual assertions and scene descriptions in the modality of abstract scenes improves performance on various textual reasoning tasks, and interpreting images in the modality of Visual Question Answering improves performance on caption retrieval, which is a visual reasoning task. With grounding, imagination and question-answering approaches to interpret images and text in different modalities, we show that learning commonsense knowledge from multiple modalities effectively improves the performance of downstream vision and language tasks, improves interpretability of the model and is able to make more efficient use of training data. Complementary to the model aspect, we also study the data aspect of commonsense learning in vision and language. We study active learning for Visual Question Answering (VQA) where a model iteratively grows its knowledge through querying informative questions about images for answers. Drawing analogies from human learning, we explore cramming (entropy), curiosity-driven (expected model change), and goal-driven (expected error reduction) active learning approaches, and propose a new goal-driven scoring function for deep VQA models under the Bayesian Neural Network framework. Once trained with a large initial training set, a deep VQA model is able to efficiently query informative question-image pairs for answers to improve itself through active learning, saving human effort on commonsense annotations.", "source_url": "https:\/\/openalex.org\/W2763796612", "pdf_url": "", "doi": "", "handle": "", "repository_name": "VTechWorks (Virginia Tech)", "source_record_id": "W2763796612", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 760, "title": "Bladder Cancer Epidemiology in Latvia and Machine Learning-Based Evaluation of Urinary Exosomal microRNAs as Diagnostic Biomarkers. Doctoral Thesis", "title_ar": "", "title_en": "Bladder Cancer Epidemiology in Latvia and Machine Learning-Based Evaluation of Urinary Exosomal microRNAs as Diagnostic Biomarkers. Doctoral Thesis", "author": "Ērika Bitiņa-Barlote", "supervisor": "", "degree_type": "phd", "degree_evidence": "نص السجل: «Doctoral Thesis»", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "predictive-analytics", "categories": "machine-learning; medical-ai; responsible-ai; predictive-analytics", "keywords": "", "abstract": "Bladder cancer (BC) is the most prevalent malignancy of the urinary system and ranks as the tenth most common cancer worldwide. Recent data indicate a significant increase in incidence, with hundreds of thousands of new cases diagnosed annually. Based on GLOBOCAN data, bladder cancer ranked 7th among cancers in Latvia in 2022, with 521 new cases. BC was the seventh most prevalent cancer in the world. The global burden of bladder cancer is considerable, with millions of individuals living with the disease. The overall global five-year prevalence was estimated at 1 700 000 in 2020. There is a substantial number of deaths attributed to it each year, making it the 13th leading cause of cancer mortality. Due to the main risk factors – smoking and chemical exposures – associated with BC, it is considered a largely preventable and avoidable cancer. BC is a heterogeneous disease with highly variable clinical, anatomic, and molecular pathology features, along with a high mutation burden. In recent years, microRNAs (miRNAs) have emerged as crucial regulators of cellular activities, and their dysregulation has been associated with the onset and progression of cancer. Although there are many studies on BC around the world, there is still no comprehensive analysis of the epidemiological trends and long-term survival rates of bladder cancer in Latvia, nor has there been an examination of the potential of exosomal microRNAs (miRNAs) as biomarkers for the diagnosis of BC. The aim of my PhD thesis was to carry out a population-based study on the epidemiological trends of bladder cancer in Latvia from 1990 to 2017, and to evaluate the diagnostic potential of urinary exosomal miRNAs in combination with clinical and laboratory data using machine learning approaches. The study also aimed at improving the understanding of cell-to-cell communication between bladder cells in patients with low-grade (LG) and high-grade (HG) BC, and the possible impact of the urinary exosomal miRNAs on the oncotransformation of normal cells. The results of the study showed that bladder cancer incidence, prevalence, and mortality increased in Latvia between 1990 and 2017. Findings revealed that the age-standardised incidence rate of bladder cancer increased from 6.8 per 100 000 in 1990 to 10.8 per 100 000 in 2017, with a statistically significant rise (APC 2.6 %) from 1990 to 2014, followed by a minor and statistically insignificant decline through 2017 (APC −3.6 %). The prevalence rates increased from 25.2 in 1990 to 69.7 in 2017, with an AAPC of 3.8 % ( 95 % CI 3.1 to 4.6) during the whole period. A statistically significant reduction was observed only in the ≥ 80-year age group, with an APC of −7.6% from 2013 to 2017. Mortality rates also increased from 3.9 to 4.4 per 100,000 during the same period, although a notable decrease was observed in the 40-59 age group, with an annual percentage change of −1.1%. Additionally, relative survival rates improved from 55.0% during 1990-2000 to 59.0% in the years 2013–2017, suggesting advancements in treatment or early detection strategies. These epidemiological findings – particularly the rising incidence and the persistent burden in older age groups – underscored the need for more sensitive, non-invasive diagnostic tools, motivating the subsequent investigation of urinary exosomal miRNAs as candidate biomarkers. Striking differences in the miRNA cargo of exosomes from the BC group as compared to the control group were identified. Among the up-regulated exosomal miRNAs, 7 miRNAs in the LG group and 3 miRNAs in the HG group were identified as directly targeting TP53. Twenty-two and 17 PTEN-targeting miRNAs were observed among the up-regulated miRNAs of the LG and HG groups, respectively. The target genes of these exosomal miRNAs and their interaction network predicted that TP53 is the strongest hub gene in both BC exosomal miRNA networks. Further, the possible role of exosomal miRNAs in synchronous BC development was discussed. Building on these differentially expressed candidates, 11 miRNAs were selected based on statistical significance and fold regulation to test whether combining these miRNAs with routine clinical and laboratory data improves diagnostic accuracy compared with using either data type alone. Machine learning was performed using the 11 specially selected miRNAs in 152 samples (89 bladder cancer patients and 63 controls), selected from an initial 210-participant cohort by applying the criterion of complete clinical data availability. Statistical analysis identified notable miRNAs, such as mir-455-3p, mir-30b-3p, and mir-429, as potential biomarkers. This study highlighted the potential of combining miRNA expression levels with laboratory tests and clinical data for improved BC diagnostics.", "source_url": "https:\/\/openalex.org\/W7225651265", "pdf_url": "", "doi": "10.25143\/prom-rsu_2026-30_dt", "handle": "", "repository_name": "", "source_record_id": "W7225651265", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 761, "title": "Bladder Cancer Epidemiology in Latvia and Machine Learning-Based Evaluation of Urinary Exosomal microRNAs as Diagnostic Biomarkers. Summary of the Doctoral Thesis", "title_ar": "", "title_en": "Bladder Cancer Epidemiology in Latvia and Machine Learning-Based Evaluation of Urinary Exosomal microRNAs as Diagnostic Biomarkers. Summary of the Doctoral Thesis", "author": "Ērika Bitiņa-Barlote", "supervisor": "", "degree_type": "phd", "degree_evidence": "نص السجل: «Doctoral Thesis»", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "predictive-analytics", "categories": "machine-learning; medical-ai; responsible-ai; predictive-analytics", "keywords": "", "abstract": "Bladder cancer (BC) is the most prevalent malignancy of the urinary system and ranks as the tenth most common cancer worldwide. Recent data indicate a significant increase in incidence, with hundreds of thousands of new cases diagnosed annually. Based on GLOBOCAN data, bladder cancer ranked 7th among cancers in Latvia in 2022, with 521 new cases. BC was the seventh most prevalent cancer in the world. The global burden of bladder cancer is considerable, with millions of individuals living with the disease. The overall global five-year prevalence was estimated at 1 700 000 in 2020. There is a substantial number of deaths attributed to it each year, making it the 13th leading cause of cancer mortality. Due to the main risk factors – smoking and chemical exposures – associated with BC, it is considered a largely preventable and avoidable cancer. BC is a heterogeneous disease with highly variable clinical, anatomic, and molecular pathology features, along with a high mutation burden. In recent years, microRNAs (miRNAs) have emerged as crucial regulators of cellular activities, and their dysregulation has been associated with the onset and progression of cancer. Although there are many studies on BC around the world, there is still no comprehensive analysis of the epidemiological trends and long-term survival rates of bladder cancer in Latvia, nor has there been an examination of the potential of exosomal microRNAs (miRNAs) as biomarkers for the diagnosis of BC. The aim of my PhD thesis was to carry out a population-based study on the epidemiological trends of bladder cancer in Latvia from 1990 to 2017, and to evaluate the diagnostic potential of urinary exosomal miRNAs in combination with clinical and laboratory data using machine learning approaches. The study also aimed at improving the understanding of cell-to-cell communication between bladder cells in patients with low-grade (LG) and high-grade (HG) BC, and the possible impact of the urinary exosomal miRNAs on the oncotransformation of normal cells. The results of the study showed that bladder cancer incidence, prevalence, and mortality increased in Latvia between 1990 and 2017. Findings revealed that the age-standardised incidence rate of bladder cancer increased from 6.8 per 100 000 in 1990 to 10.8 per 100 000 in 2017, with a statistically significant rise (APC 2.6 %) from 1990 to 2014, followed by a minor and statistically insignificant decline through 2017 (APC −3.6 %). The prevalence rates increased from 25.2 in 1990 to 69.7 in 2017, with an AAPC of 3.8 % ( 95 % CI 3.1 to 4.6) during the whole period. A statistically significant reduction was observed only in the ≥ 80-year age group, with an APC of −7.6% from 2013 to 2017. Mortality rates also increased from 3.9 to 4.4 per 100,000 during the same period, although a notable decrease was observed in the 40-59 age group, with an annual percentage change of −1.1%. Additionally, relative survival rates improved from 55.0% during 1990-2000 to 59.0% in the years 2013–2017, suggesting advancements in treatment or early detection strategies. These epidemiological findings – particularly the rising incidence and the persistent burden in older age groups – underscored the need for more sensitive, non-invasive diagnostic tools, motivating the subsequent investigation of urinary exosomal miRNAs as candidate biomarkers. Striking differences in the miRNA cargo of exosomes from the BC group as compared to the control group were identified. Among the up-regulated exosomal miRNAs, 7 miRNAs in the LG group and 3 miRNAs in the HG group were identified as directly targeting TP53. Twenty-two and 17 PTEN-targeting miRNAs were observed among the up-regulated miRNAs of the LG and HG groups, respectively. The target genes of these exosomal miRNAs and their interaction network predicted that TP53 is the strongest hub gene in both BC exosomal miRNA networks. Further, the possible role of exosomal miRNAs in synchronous BC development was discussed. Building on these differentially expressed candidates, 11 miRNAs were selected based on statistical significance and fold regulation to test whether combining these miRNAs with routine clinical and laboratory data improves diagnostic accuracy compared with using either data type alone. Machine learning was performed using the 11 specially selected miRNAs in 152 samples (89 bladder cancer patients and 63 controls), selected from an initial 210-participant cohort by applying the criterion of complete clinical data availability. Statistical analysis identified notable miRNAs, such as mir-455-3p, mir-30b-3p, and mir-429, as potential biomarkers. This study highlighted the potential of combining miRNA expression levels with laboratory tests and clinical data for improved BC diagnostics.", "source_url": "https:\/\/openalex.org\/W7225633027", "pdf_url": "", "doi": "10.25143\/prom-rsu_2026-30_dts", "handle": "", "repository_name": "", "source_record_id": "W7225633027", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 763, "title": "Non-invasive fairness drift monitor for machine learning models using conformance constraints: a thesis in Computer Science", "title_ar": "", "title_en": "Non-invasive fairness drift monitor for machine learning models using conformance constraints: a thesis in Computer Science", "author": "Raquel Ana Magalhães Bush", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; computer-vision; responsible-ai; object-detection; fairness", "keywords": "", "abstract": "Machine learning models deployed in consequential domains can become unfair toward protected subgroups as the data they receive drifts over time, yet the protected attributes needed to measure fairness directly are often unavailable at runtime due to privacy regulation and operational constraints. This creates a gap: existing fairness toolkits require protected labels and perform one-time audits, while generic drift detectors monitor continuously but cannot localize which subgroup a shift harms. This thesis develops a non-invasive fairness drift monitor that addresses this gap by repurposing Conformance Constraints, a data-profiling primitive, as a temporal fairness signal. The monitor learns per-subgroup distributional profiles at baseline, using protected attributes only once, and thereafter tracks violation of those profiles over incoming data batches without any runtime access to protected attributes. Across three fairness benchmarks, two classifiers, and nineteen controlled drift scenarios, a minority-subgroup conformance-constraint signal tracks fairness degradation more closely than KS and KL detectors under global drift on the two datasets where it is defined, while the conformance constraint signals defined on all three datasets remain competitive with those baselines under both regimes. Subgroup localization provides its clearest advantage under global drift, where a minority-subgroup signal substantially outperforms both aggregate signals and the baselines. The correlations are modest in absolute terms, indicating that the monitor functions as a screening instrument that flags fairness degradation for closer investigation rather than as a precise estimator. The approach offers privacy-preserving, subgroup-aware fairness monitoring suited to regulated deployment settings.", "source_url": "https:\/\/openalex.org\/W7212174866", "pdf_url": "", "doi": "10.62791\/20639", "handle": "", "repository_name": "", "source_record_id": "W7212174866", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 765, "title": "Emulating marine ecosystem dynamics: a machine learning approach for biomass prediction and forecasting : a thesis in Data Science", "title_ar": "", "title_en": "Emulating marine ecosystem dynamics: a machine learning approach for biomass prediction and forecasting : a thesis in Data Science", "author": "Ajmal Abbas", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "deep-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; deep-learning; predictive-analytics", "keywords": "", "abstract": "Marine ecosystem models such as Atlantis have valuable insight into multispecies interactions, spatial dynamics and fisheries management, but their high computational cost limits are rapid scenario analysis and real-time decision making. This study presented a data-driven ecosystem emulator for the Northeast U.S. Atlantis model using automated machine learning approaches. A large-scale dataset (~3.4 million records) spanning 1964-2020 was contracted, integrating biomass of species functional groups, spatial polygons temporal indices and environments variables such as temperature and salinity. The emulator framework employed automated machine learning techniques, including Random Forest and Extra tress regression, with model selection and hyperparameter optimization performance using Automatic Machine Learning strategies. In addition, Autokeras was utilized to explore neural network architecture in an automated manner, enabling data-driven model allocation, lagged variables to capture ecological inertia and time-aware transformations. Model performance was evaluated using out-of-sample temporal validation, recursive back testing and ecological plausibility assessments. Result demonstrated string predictive performance. with species-level R2 values frequently exceeding 0.90 and overall model accuracy approaching 94%. The emulator achieved high computational efficiency, with end-to-end prediction completed in under few seconds, substantially reducing runtime compared to Atlantis simulations. This work established a scalable and efficient AutoML-driven alternative to process -based ecosystem, model, enabling rapid biomass estimation and supporting data-driven fisheries management and ecosystem analysis.", "source_url": "https:\/\/openalex.org\/W7167819629", "pdf_url": "", "doi": "10.62791\/20585", "handle": "", "repository_name": "", "source_record_id": "W7167819629", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 771, "title": "Machine learning with FPGA acceleration for signal classification: a thesis in Computer Engineering", "title_ar": "", "title_en": "Machine learning with FPGA acceleration for signal classification: a thesis in Computer Engineering", "author": "Bruno Paiva", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "computer-vision", "research_subdomain": "cnn", "categories": "machine-learning; deep-learning; computer-vision; cnn", "keywords": "", "abstract": "Spectrum sharing in the next generation (NextG) wireless communication systems can solve the spectrum scarcity problem, meet the demands of high data rate, and provide better quality of service. Characterizing surrounding wireless signals is essential to support such a spectrum sharing over the same spectrum band. Using machine learning to identify different signals on the same channel can solve this issue and lead to real time signal de-modulation of an unknown signal class. Using existing machine learning techniques such as convolution neural networks (CNN), Region-based convolution neural networks (R-CNN),fast region-based convolution neural networks (Fast R-CNN) and faster region-based convolution neural networks (Faster R-CNN), one or more signals can be classified and extracted from a channel at the same time. This project evaluates the image pre-processing methods to better format signal data for machine learning, including machine learning with time do-main signals, using spectrograms which maps frequency and time components of the signal, and scalograms which maps the continuous wavelet transform to time. These models are then built in TensorFlow and compiled to run on a CPU, GPU, and a Xilinix FPGA usingVitis-AI, to evaluate models for accuracy and inference time. After evaluating the models, it is found that the spectrogram frames are not an ideal image pre-processing technique for this application, with possible upside with higher frequency signals. The continuous wavelet transform performed in the scalogram frame amounts to an excellent prepossessing technique. Region-based convolution neural networks prove to offer a reduction in classification time with a significant reduction in classification accuracy, A better method for reducing classification time is FPGA acceleration. As it provides a tenfold increase in classification speed while retaining the exact same accuracy when compared to the same neural network running on a mid-range GPU.", "source_url": "https:\/\/openalex.org\/W4409526988", "pdf_url": "", "doi": "10.62791\/20255", "handle": "", "repository_name": "", "source_record_id": "W4409526988", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 774, "title": "Machine learning-based angle-of-arrival estimation of RF signals: a thesis in Electrical Engineering", "title_ar": "", "title_en": "Machine learning-based angle-of-arrival estimation of RF signals: a thesis in Electrical Engineering", "author": "Jin Feng Lin", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "deep-learning", "research_subdomain": "cnn", "categories": "machine-learning; deep-learning; cnn", "keywords": "", "abstract": "This thesis presents a comprehensive study on angle of arrival (AoA) estimation techniques for spectrum sharing in next-generation (NextG) wireless communication systems. The investigation encompasses signal processing methods, including multiple signal classification (MUSIC) and estimation of signal parameters via rotational invariance techniques (ESPRIT), as well as machine learning (ML) techniques such as artificial neural networks (ANNs) and convolutional neural networks (CNNs). These techniques are evaluated using both synthetic and over-the-air (OTA) test scenarios, with a focus on the impact of multipath fading and varying signal-to-noise-ratio (SNR) conditions. The study investigates the role of ML algorithms in enhancing the performance of wireless communication systems under challenging conditions. The robustness of ML techniques in real-world scenarios, including their abilities to maintain stability under multipath fading, is highlighted. The thesis proposes a new ML algorithm aimed at optimizing the performance of providing positional awareness about radio frequency (RF) signals. Furthermore, the finding with testing with OTA tests, reveal that while the MUSIC algorithm excels in synthetic scenarios with high-resolution capabilities beyond the antenna array’s resolution, it suffers from performance degradation in non-line-of-sight (NLOS) conditions and low SNR environments. On the other hand, ML algorithms demonstrate robustness and stability across varying conditions, though they exhibit a slightly higher error rate in simulations.", "source_url": "https:\/\/openalex.org\/W4409785432", "pdf_url": "", "doi": "10.62791\/20367", "handle": "", "repository_name": "", "source_record_id": "W4409785432", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 780, "title": "Personalized Support Vector Machines for Privacy-Preserving Federated Learning", "title_ar": "", "title_en": "Personalized Support Vector Machines for Privacy-Preserving Federated Learning", "author": "Aleksei Ponomarenko-Timofeev", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Tampere University", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "computer-vision", "research_subdomain": "federated-learning", "categories": "machine-learning; computer-vision; responsible-ai; ai-applications; predictive-analytics; federated-learning; fairness; ai-cybersecurity", "keywords": "", "abstract": "The pervasive digitalization of the environment enables monitoring of real-world systems by collecting data from multiple diverse sources. Such data may correspond to problems such as network intrusion detection or predicting the state of the wireless channel. While machine learning (ML)-based solutions are increasingly employed in such tasks, centralized learning approaches often fall short in terms of scalability and privacy. A more practical alternative is offered by federated learning (FL), which enables collaborative model training without disclosing raw data. However, FL brings a new set of challenges such as heterogeneous data distributions, diverse computational resources, and privacy requirements, which are rarely considered together in the algorithm design. This dissertation addresses the heterogeneity of the data structure and distribution, the heterogeneity of the computational resources, and the requirements for privacy preservation. Motivated by the fact that such factors may be present simultaneously, this dissertation focuses on developing lightweight privacy-preserving personalized FL algorithms. The work aims to bridge the gap in the FL field by formulating a set of algorithms for learning classification\/regression models (i), proposing policies for operating in highly heterogeneous systems and assessing their applicability in different scenarios (ii), assessing the applicability of conventional privacy preservation measures (iii), and proposing novel privacy-preserving mechanisms and evaluating their applicability (iv). Assessments are based on the results collected from extensive experiments across a broad range of scenarios and data types. The numerical results demonstrate the effectiveness of the proposed methods. In systems with heterogeneous computational resources and data, the proposed algorithms consistently maintain performance on par with or better than the state-of-the-art. The training delay was observed to be up to 20 % lower compared with the state-of-the-art personalized FL algorithms. Moreover, the proposed policies for operating in highly heterogeneous systems retain fairness and leverage heterogeneity to improve the quality of the learned models. The positive effect of these policies was observed through faster model convergence and stable fairness. It was also determined that conventional privacy preservation measures, such as differential privacy mechanisms, may be applied to the system where the proposed algorithms are used. The utility of the models learned by the algorithms was assessed for these privacy preservation measures. The proposed privacy preservation mechanisms, based on multiplicative noise, were found to provide better privacy\/utility trade-offs compared to the conventional Gaussian mechanism within a certain operational region. These regions of higher utility were explored for diverse datasets (e.g., image-based, time-series). Multiple approaches to improving the utility of the learned model while retaining the privacy guarantees were studied based on the properties of the algorithms. The advantages of these approaches are demonstrated through both qualitative insights and quantitative evidence. Overall, this thesis addresses a key gap in the FL literature by providing lightweight, privacy-preserving, and personalized algorithms designed for heterogeneous wireless systems.", "source_url": "https:\/\/openalex.org\/W7132950646", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Tampere University Institutional Repository (Tampere University)", "source_record_id": "W7132950646", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 781, "title": "Securing Federated Learning: A Comprehensive Defence Against Privacy Attacks", "title_ar": "", "title_en": "Securing Federated Learning: A Comprehensive Defence Against Privacy Attacks", "author": "Moyeen, M A", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Concordia University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "federated-learning", "categories": "machine-learning; medical-ai; reinforcement-learning; responsible-ai; ai-applications; federated-learning; fairness; ai-cybersecurity; ai-industry", "keywords": "", "abstract": "In this information age, machine learning (ML) applications drive smart living through innovations such as personalized healthcare, intelligent transportation, and smart homes. However, despite these advancements, businesses and industries continue to face significant challenges in safeguarding data privacy, as ML systems increasingly rely on vast amounts of user data. In this direction, Federated Learning (FL) has emerged as a promising solution, enabling collaborative model training while keeping user data on local premises without the need to share raw data. However, FL faces significant challenges also, including expensive communication costs, system heterogeneity, and vulnerability to various attacks. In particular, it is susceptible to poisoning attacks, where malicious participants corrupt models and data as well as inference attacks that exploit gradients to reveal sensitive information through membership inference or model inversion techniques. These attacks can extract sensitive information from shared gradients, undermining the fundamental privacy guarantees of federated systems. Thus, effective defence mechanisms are fundamental for fully leveraging the advantages of FL. Numerous defences, such as FoolsGold, Flod, Flad, MADDPG, and others, are in place to secure the FL systems. However, the majority of the defence mechanisms suffer from accuracy degradation, computational overhead, and inadequate attack prevention. Most client selection methods cannot reliably separate malicious and straggler clients, with even cutting-edge approaches struggling with herding and cold-start issues. Furthermore, recent state-of-the-art techniques frequently fail to defend against inference attacks adequately. These methods typically employ Secure Multiparty Computation (SMPC), Homomorphic Encryption (HE), or Differential Privacy (DP) as defensive measures. However, SMPC and HE suffer from high computational complexity, while DP often leads to degraded model accuracy. Thus, this research addresses these limitations by proposing five different defence mechanisms that ensure robust FL with protected gradients. The proposed mechanisms consist of FedChallenger, Fed-Reputed, SignDefence, Ada-Sign, and SignMPC. The proposed FedChallenger introduces a dual-layer defence mechanism that comprises the zero-trust challenge-response-based authentication at the first layer and a variant of Trimmed-Mean aggregation at the second layer that leverages pairwise cosine similarity and Median Absolute Deviation (MAD). Extensive evaluation on MNIST, FMNIST, EMNIST, and CIFAR-10 datasets demonstrates 3-10% accuracy improvement over state-of-the-art approaches with 1.1-2.2 times faster convergence and 2-3% higher F1-scores. Subsequently, the reputation-based client selection approach, Fed-Reputed, leverages device capability information and a modified Bellman equation within a hierarchical framework, integrated into a Deep Q-Learning Network (DQN)-based Imbalanced Classification Markov Decision Process (ICMDP) classifier for enhanced client selection. Testing on MNIST and FMNIST datasets demonstrates 9-50% accuracy gains and 1.3-1.7 times faster convergence while effectively detecting both malicious and straggler clients. Moreover, existing methods often suffer from the dying ReLU problem, where neurons permanently deactivate during training. To counter the dying ReLU problem, SignDefence implements a sophisticated aggregation scheme that utilizes sign direction and LeakyReLU-based aggregation, incorporating Jaccard similarity derived from binary-encoded model weights. This technique demonstrates consistent accuracy and F1-score improvements across different attack conditions. Despite its benefits, SignDefence remains vulnerable to inference attacks and suffers from limited generalization due to its fixed threshold across diverse benchmark datasets. To address these limitations, a lightweight strategy, Ada-Sign, employs adaptive threshold computation and incorporates DP mechanisms. This approach maintains comparable accuracy to SignDefence while providing enhanced gradient protection through adaptive DP. Extensive evaluation on MNIST and HAR datasets reveals 3-20% accuracy improvement for Ada-Sign over the majority of state-of-the-art techniques. Finally, to enhance the protection of both SignDefence and Ada-Sign against inference attacks, iv SignMPC integrates highly configurable SMPC, HE, and DP algorithms. This combined approach ensures comprehensive communication security and gradient privacy while avoiding significant performance bottlenecks. Comprehensive evaluation on MNIST and HAR datasets demonstrates 4-17% accuracy gains for SignMPC over established approaches while maintaining computational efficiency and robust privacy guarantees.", "source_url": "https:\/\/openalex.org\/W7112513152", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Spectrum Research Repository (Concordia University)", "source_record_id": "W7112513152", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 784, "title": "Privacy preserving vertically partitioned federated learning: new techniques and considerations", "title_ar": "", "title_en": "Privacy preserving vertically partitioned federated learning: new techniques and considerations", "author": "Florian van Daalen", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Maastricht University", "university_source": "repository_host", "college": "", "department": "", "country": "Netherlands", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "federated-learning", "categories": "medical-ai; responsible-ai; federated-learning; fairness", "keywords": "", "abstract": "Research depends on the collection of large amounts of data. However, this data is collected by different institutes, and is often difficult to share because of privacy concerns. This thesis proposes a number of new techniques that can be used to run data-analysis on data in a privacy preserving manner. It uses federated learning techniques in line with the “personal health train”-approach proposed by Dr. Andre Dekker. This thesis was part of the CARRIER project and focuses on data-sharing scenarios that are “vertically” partitioned. In such a scenario the parties sharing data have collected different types of data on the same population. For example, in the CARRIER data the data is split in a socio-economic dataset, collected by Statistics Netherlands, and medical data, collected at the MUMC and by regional GPs. All parties are collected data from the south-Limburg population. In addition to proposing new techniques, this thesis also discusses the current approach to privacy in research. It critiques the influence of large institutes, and questions if we are really protecting the data-subjects.", "source_url": "https:\/\/openalex.org\/W4408328439", "pdf_url": "", "doi": "10.26481\/dis.20250328fd", "handle": "", "repository_name": "Research Publications (Maastricht University)", "source_record_id": "W4408328439", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 788, "title": "Enhancing Privacy and Efficiency in Federated Learning: Quantization and Low-Rank Adaptation Approaches", "title_ar": "", "title_en": "Enhancing Privacy and Efficiency in Federated Learning: Quantization and Low-Rank Adaptation Approaches", "author": "Tianqu Kang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "responsible-ai", "research_subdomain": "federated-learning", "categories": "machine-learning; responsible-ai; ai-applications; transfer-learning; federated-learning; fairness; ai-cybersecurity", "keywords": "", "abstract": "Federated Learning (FL) enables collaborative AI model training on decentralized data but faces challenges in ensuring privacy and operational efficiency, especially for large models over wireless networks. This thesis addresses these critical issues by enhancing both aspects in FL systems. Firstly, we provide a rigorous analysis of the interplay between model update quantization and Differential Privacy (DP). Using Rényi Differential Privacy (RDP), we formally demonstrate that lower quantization bit levels can enhance privacy guarantees when applied to Gaussian-noised updates. This theoretical finding, contrary to some prior notions, is empirically validated using Membership Inference Attacks (MIA), confirming that quantization can be a privacy-enhancing mechanism in addition to a communication-saving one. Secondly, we propose a novel framework for differentially private federated fine-tuning (FedFT) of Foundation Models using Low-Rank Adaptation (LoRA). This framework uniquely leverages inherent wireless channel noise as a natural source for DP, reducing the need for artificial noise injection. We analyze the convergence of this split FedFT-LoRA architecture and address the critical problem of noise amplification by proposing a modified LoRA update strategy where only one low-rank matrix is trained while another initialized as scaled orthogonal. Simulations show this approach achieves superior model accuracy under stringent privacy budgets compared to baseline methods. Collectively, this work offers new theoretical insights and practical solutions for deploying private and resource-efficient FL, particularly for large-scale models in wireless environments.", "source_url": "https:\/\/openalex.org\/W4414315150", "pdf_url": "", "doi": "10.14711\/thesis-hdl152373", "handle": "", "repository_name": "", "source_record_id": "W4414315150", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 789, "title": "Secure and Privacy-preserving Federated Learning with Explainable Artificial Intelligence for Smart Healthcare System", "title_ar": "", "title_en": "Secure and Privacy-preserving Federated Learning with Explainable Artificial Intelligence for Smart Healthcare System", "author": "Ali Raza", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "medical-ai", "research_subdomain": "federated-learning", "categories": "machine-learning; deep-learning; medical-ai; responsible-ai; ai-applications; xai; federated-learning; fairness; ai-cybersecurity", "keywords": "", "abstract": "The growing population around the globe has a significant impact on various sectors including the labor force, healthcare, and the global economy. The healthcare sector is among the most affected sectors by the growing population due to the increasing demand for resources including doctors, nurses, equipment, and healthcare facilities. Intelligent systems have been incorporated to enhance decision-making, management, prognosis, and diagnosis in order to tackle such issues and offer improved healthcare to patients. Among such systems, those based on deep learning (DL), a subclass of machine learning (ML) have outperformed many traditional statistical and ML systems owing to their capability of automatically discovering and learning related features for a given task and robustness. Therefore, the use of DL has seen a steady increase in many applications. Nevertheless, usually, the training of DL models relies on a single centralized server, which brings many challenges: (1) except for some big enterprises most of the small enterprises have limited quality data, which is insufficient to support the training of data-hungry DL models, (2) access to data, which is vital for these systems, often raises privacy concerns. The collection and analysis of sensitive patient information must be done in a secure and ethical manner to ensure the protection of individual privacy rights, (3) high communication cost and computation resources required, (4) a large number of trainable parameters make the outcome of DL hard to explain, which is required in some applications, such as healthcare. Compared to centralized ML, federated learning (FL) improves both privacy and communication costs, where clients collaboratively train a joint model without sharing the raw data directly. FL minimizes privacy breaches and safeguards sensitive data by keeping it distributed locally. This enables collaborative model training while reducing the risk of unauthorized access and data breaches. Additionally, it promotes data diversity and scalability by involving multiple sources in joint model training and decreases communication costs by sharing only model updates instead of the entire dataset. However, FL brings its own challenges. For example, heterogeneous local data among the clients makes it challenging to train a high-performing and robust global model. Sharing updates (hundreds of thousands of parameters) still has high communication costs. Additionally, the distributed nature and access control of local data in FL make it more vulnerable to malicious attacks. Moreover, the challenge of explaining the results of DL still remains challenging, and methods are needed to be developed to bring trust accountability, and transparency in sensitive applications, such as healthcare. Therefore, the aim of this thesis is to create robust frameworks that are secure, high-performing, and privacy-friendly within federated settings. These frameworks will be specifically designed for end-to-end (we train our frameworks using raw data without any manual feature extraction) healthcare applications, considering the presence of non-identically distributed data among clients in FL to bring robustness. By addressing these challenges, the objective is to enhance the overall system's resilience and effectiveness. We also propose a methodology for detecting anomalies within federated settings, particularly in applications with limited available data for the abnormal class. Furthermore, clients in FL are usually resource-constrained with limited computation and communication resources available. Therefore, to support efficient computation and communication in a federated setting we propose a lightweight framework (in terms of the trainable number of parameters). Additionally, to provide explanations of the DL models' outcomes, which are usually hard to explain because of the large number of parameters, we propose model-agnostic explainable AI modules to help explain the results of DL models. Moreover, in order to protect the proposed frameworks against cyber attacks, such as poisoning attacks, we propose a framework in federated settings, which makes the proposed healthcare frameworks more secure and trustworthy. Finally, with experimental analysis using baseline datasets for one of the most common health conditions i.e., cardiovascular diseases (arrhythmia detection, ECG anomaly detection) and human activity recognition (used for supplementing cardiovascular diseases detection), we show the superiority of the proposed frameworks over state-of-the-art work.", "source_url": "https:\/\/openalex.org\/W4399027046", "pdf_url": "http:\/\/www.theses.fr\/2023ULILB019\/document", "doi": "10.22024\/unikent\/01.02.102717", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W4399027046", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 793, "title": "Addressing data heterogeneity, scarcity, and training efficiency in privacy-preserving federated learning", "title_ar": "", "title_en": "Addressing data heterogeneity, scarcity, and training efficiency in privacy-preserving federated learning", "author": "Chen, Haokun", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Ludwig-Maximilians-Universität München", "university_source": "repository_host", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; generative-ai; medical-ai; responsible-ai; ai-applications; transfer-learning; few-zero-shot; optimization; multimodal; llm; diffusion; federated-learning; fairness; ai-industry", "keywords": "", "abstract": "Over the past decade, deep learning has achieved significant breakthroughs in various domains, including computer vision and natural language processing, with applications spanning industries such as healthcare and manufacturing. However, the success of deep neural networks (DNNs) relies heavily on access to large-scale datasets, which can be difficult for individual organizations to acquire. The reasons behind are the complexities and costs for data collection and annotation at large scale. A straightforward solution is to centralize data from multiple organizations for training. However, this approach raises significant privacy concerns, as such data often contains sensitive or confidential information. Furthermore, regulations like the General Data Protection Regulation (GDPR) emphasize the importance of protecting user privacy in inter-organizational data exchanges. Moreover, transmitting large volumes of data introduces substantial computational overhead, further complicating the process. These challenges highlight the urgent need for methods that facilitate collaborative data use while ensuring privacy preservation. Federated learning (FL), which enables multiple parties to collaboratively train a DNN with the assistance of a central server, offers an effective solution to the aforementioned problem. Unlike traditional centralized learning, which requires collecting data from each party, FL eliminates the need to upload data for joint training. Instead, locally trained models are exchanged with a central server, which aggregates the knowledge from all of the uploaded models and then distributes a refined global model to each party. This approach allows each party to benefit from the collective contributions, ultimately enhancing the model performance. This thesis makes four key contributions to addressing challenges in federated learning, including data heterogeneity, data scarcity, and system convergence rate, while preserving client data privacy. For each contribution, we propose a novel method and empirically demonstrate its effectiveness within the relevant problem context. In this thesis, we address key challenges in federated learning (FL) related to data heterogeneity, foundation model adaptation, hyperparameter tuning, and communication efficiency. First, we tackle feature space heterogeneity by introducing a generative augmentation method that aligns diverse client distributions, validated on both public and real-world datasets. We then adapt parameter-efficient fine-tuning techniques for vision-language models in FL by designing a dual-stream adapter that captures both client-specific and client-agnostic knowledge. To optimize performance under limited resources and non-IID conditions, we propose an evolutionary hyperparameter tuning framework that enables efficient online optimization. Lastly, we address data scarcity and heterogeneity in One-Shot FL by personalizing pretrained latent diffusion models, enabling privacy-preserving synthetic data generation that improves performance across challenging domains such as medical and satellite imaging.", "source_url": "https:\/\/openalex.org\/W7108654185", "pdf_url": "", "doi": "10.5282\/edoc.36097", "handle": "", "repository_name": "Electronic Theses of LMU Munich (Ludwig-Maximilians-Universität München)", "source_record_id": "W7108654185", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 794, "title": "Federated Learning-Based Intrusion Detection System for Privacy-Preserving IoT Networks", "title_ar": "", "title_en": "Federated Learning-Based Intrusion Detection System for Privacy-Preserving IoT Networks", "author": "Prasanna Dhanireddy ; Khushboo Kumari", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "KTH Royal Institute of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "deep-learning; medical-ai; responsible-ai; ai-applications; cnn; transformers; federated-learning; fairness; ai-cybersecurity; ai-industry", "keywords": "", "abstract": "The rapid growth of IoT devices in smart homes, healthcare, and industrial systems has increased vulnerability to cyber threats. Traditional centralized intrusion detection systems (IDS) raise privacy, and scalability concerns due to the need for centralized data collection. This project aims to develop a Federated Learning (FL)-based IDS that enables distributed IoT nodes to collaboratively train the models without sharing raw data. This study proposes a FL-based IDS framework for IoT environments that systematically evaluates the trade-offs between detection performance, privacy guarantees, model efficiency, and synthetic zero-day like attack robustness. The framework investigates and compares multiple deep learning architectures, including Multilayer Perceptron (MLP), one-dimensional Convolutional Neural Networks (1D-CNN), and TabTransformer (TTF) architectures. The integration of differential privacy was also examined to enhance data protection and analyze its impact on model performance. Additionally, quantization techniques were investigated for the models to reduce model size and improve computational efficiency for deployment on resource-constrained IoT devices. The proposed system is evaluated using the benchmark dataset CICIoT2023 basedon standard performance metrics such as accuracy, precision, recall, F1-score, AUC-ROC score along with inference time, and communication overhead. Experimental results demonstrate that TTF achieved the highest detection performance with an accuracy of 93.97%, F1-score of 92.47%, and AUC-ROC score of 98.11%, while MLP provided superior computational efficiency for resource-constrained IoT devices. The integration of differential privacy introduced a privacy-utility trade-off, where model accuracy decreased from 93.97% to 91.40% under stricter privacy guarantees. Furthermore, post-training dynamic quantization reduced TTF and 1D-CNN model size by 67.72% and 65.66%, thereby decreasing inference time with minimal performance degradation. The optimized federated model also demonstrated robustness against synthetic zero-day like attack scenarios generated using CTGAN. It achieved an accuracy of 90.96%, F1-score of 90.16%, and AUC-ROC of 96.55%, suggesting its potential to detect diverse attack behaviors.", "source_url": "https:\/\/openalex.org\/W7213389565", "pdf_url": "", "doi": "", "handle": "", "repository_name": "KTH Publication Database DiVA (KTH Royal Institute of Technology)", "source_record_id": "W7213389565", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 805, "title": "An optimized internet of vehicles network for safety-based messages transmission using particle swarm optimization", "title_ar": "", "title_en": "An optimized internet of vehicles network for safety-based messages transmission using particle swarm optimization", "author": "Shahirah Mohamed Hatim", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Universiti Teknologi MARA", "university_source": "repository_host", "college": "", "department": "", "country": "Malaysia", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "machine-learning", "research_subdomain": "optimization", "categories": "machine-learning; ai-applications; optimization; ai-smart-cities", "keywords": "", "abstract": "This thesis addresses the problem of traffic congestion, which arises from increased vehicular queuing when the packet transmission load in the network exceeds the available buffer capacity. This condition leads to slower travel speed and also delays in travel time. Many urban cities worldwide have experienced severe congestion challenges. However, with the aid of the Internet of Things (IoT), the traffic congestion problem in the event can be managed effectively as vehicles are now equipped with sensors, actuators, data acquisition and communication systems that enable inter-vehicle communication. The Internet of Vehicles (IoV) has emerged as a key mechanism in in managing traffic congestion within modern Intelligent Transportation Systems (ITS), providing extensive opportunities for safety and mobility applications. This networking technology, provides vehicles with endless possibilities of applications including safety-related applications. Due to the high speed, dynamic mobility and unstable connectivity of vehicular nodes, maintaining stable IoV network performance remains a challenge. Previous research has explored various congestion control optimization techniques such Genetic Algorithm, Ant Colony and Ant Bee Colony for vehicular networks. While these methods improved performance to some extent, they often suffer from high computational complexity or slow convergence, limiting their suitability for real-time IoV environments. The recent proliferation of IoV network has emphasized the need for high Quality of Service (QoS) requirements in all networking scenarios. This study provides a comprehensive evaluation of the impact of vehicular mobility on IoV network performance. The analysis focuses on key QoS metrics, namely throughput, packet delivery ratio (PDR) and delay, while examining the relationship between mobility factors and medium access behaviour under varying network conditions. In achieving better performance in the QoS of the IoV network, a bio-inspired optimization approach based on the Particle Swarm Optimization (PSO) algorithm is proposed to mitigate network performance degradation caused by high node mobility. Vehicle mobility data is obtained from traffic flow simulation in SUMO and the network simulation phase is conducted in OMNeT++. Safety messages are disseminated among vehicles within the network. The experiment for the IoV network is concluded by optimizing the transmission rate to minimize the delay as the objective function in PSO algorithm. To ensure stable performance evaluation under dynamic vehicular conditions, simulations were conducted using varying simulation times and numbers of vehicles to capture both temporal network behaviour and traffic density variations. Under these settings, the throughput of the IoV network recorded average improvements of up to 23% and 14% after optimization. Packet delivery ratio (PDR) was also enhanced, with average gains of up to 37% and 15%, while network delay was reduced by up to 12% and 15% across different simulation times and vehicle densities. These results highlight the effectiveness of the optimization in improving IoV network performance under diverse conditions. The proposed PSO-based approach introduces a dynamic transmission rate control mechanism that adapts to vehicular density, effectively reducing congestion on the control channel. Overall, the findings highlight the potential for future research and practical applications in vehicular communication systems.", "source_url": "https:\/\/openalex.org\/W7170115999", "pdf_url": "https:\/\/ir.uitm.edu.my\/id\/eprint\/142641\/1\/142641_fulltext.pdf", "doi": "", "handle": "", "repository_name": "UiTM Institutional Repositories (Universiti Teknologi MARA)", "source_record_id": "W7170115999", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 810, "title": "Short-Term Load Forecasting Using A Hybrid Of Genetic Algorithm (Ga) And Particle Swarm Optimization (Pso) For An Optimized Neural Network", "title_ar": "", "title_en": "Short-Term Load Forecasting Using A Hybrid Of Genetic Algorithm (Ga) And Particle Swarm Optimization (Pso) For An Optimized Neural Network", "author": "Stephen T Siri", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Nairobi", "university_source": "repository_host", "college": "", "department": "", "country": "Kenya", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "deep-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; deep-learning; predictive-analytics; optimization", "keywords": "", "abstract": "Short-term load forecasting (STLF) has emerged as one of the most important fields of study for power system operation for system efficiency and reliability. It plays a significant role in load flow analysis, contingency analysis, planning, scheduling and maintenance of power systems facilities; therefore, the system cost-effectiveness is determined by accurate load forecast. Numerous researchers have been done to improve the accuracy of the conventional methods such as time series, regression analysis or autoregressive moving average (ARMA) and the use of Artificial Neural Networks (ANNs) in load forecasting. ANN has shown more accurate results than the others. But the training of ANNs, with a back-propagation algorithm or gradient algorithms, requires long processing time has the difficulty in selecting the optimal order of the components and trapping in local minima. This research aimed at solving this problem by proposing a hybrid method based on Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) for training and optimizing the weights of ANN. The proposed hybrid method enables a reduction in the search space and the iteration time. The proposed algorithm was tested in MATLAB 2016® software using 24 hourly load data of different days (i.e. weekdays and weekends) from Juba Power Plant (JPP), South Sudan. PSO, GA and a hybrid of genetic algorithm with particle swarm optimization (HGAPSO) and ANN were studied and the resulting mean absolute percentage error (MAPE) found to be range from 1.9% to 3.40%, 2.23% to 3.65% and 1.47% to 1.98% respectively. The results obtained were compared and it was observed that HGAPSO-ANN method has a better performance in reducing and improving forecast error compared to PSO-ANN and GA-ANN methods. Therefore, a hybridized HGAPSO algorithm with ANN improves forecast accuracy.", "source_url": "https:\/\/openalex.org\/W2936546245", "pdf_url": "", "doi": "", "handle": "", "repository_name": "University of Nairobi Research Archive (University of Nairobi)", "source_record_id": "W2936546245", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 821, "title": "Using Particle Swarm Optimization for Market Timing Strategies", "title_ar": "", "title_en": "Using Particle Swarm Optimization for Market Timing Strategies", "author": "Ismail Mohamed", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Kent", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "machine-learning", "research_subdomain": "optimization", "categories": "machine-learning; optimization", "keywords": "", "abstract": "Market timing is the issue of deciding when to buy or sell a given asset on the market. As one of the core issues of algorithmic trading systems, designers of such system have turned to computational intelligence methods to aid them in this task. In this thesis, we explore the use of Particle Swarm Optimization (PSO) within the domain of market timing.nPSO is a search metaheuristic that was first introduced in 1995 [28] and is based on the behavior of birds in flight. Since its inception, the PSO metaheuristic has seen extensions to adapt it to a variety of problems including single objective optimization, multiobjective optimization, niching and dynamic optimization problems. Although popular in other domains, PSO has seen limited application to the issue of market timing. The current incumbent algorithm within the market timing domain is Genetic Algorithms (GA), based on the volume of publications as noted in [40] and [84]. In this thesis, we use PSO to compose market timing strategies using technical analysis indicators. Our first contribution is to use a formulation that considers both the selection of components and the tuning of their parameters in a simultaneous manner, and approach market timing as a single objective optimization problem. Current approaches only considers one of those aspects at a time: either selecting from a set of components with fixed values for their parameters or tuning the parameters of a preset selection of components. Our second contribution is proposing a novel training and testing methodology that explicitly exposes candidate market timing strategies to numerous price trends to reduce the likelihood of overfitting to a particular trend and give a better approximation of performance under various market conditions. Our final contribution is to consider market timing as a multiobjective optimization problem, optimizing five financial metrics and comparing the performance of our PSO variants against a well established multiobjective optimization algorithm. These algorithms address unexplored research areas in the context of PSO algorithms to the best of our knowledge, and are therefore original contributions. The computational results over a range of datasets shows that the proposed PSO algorithms are competitive to GAs using the same formulation. Additionally, the multiobjective variant of our PSO algorithm achieve statistically significant improvements over NSGA-II.", "source_url": "https:\/\/openalex.org\/W3136655786", "pdf_url": "", "doi": "10.22024\/unikent\/01.02.87076", "handle": "", "repository_name": "Kent Academic Repository (University of Kent)", "source_record_id": "W3136655786", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 823, "title": "An improved data classification framework based on fractional particle swarm optimization", "title_ar": "", "title_en": "An improved data classification framework based on fractional particle swarm optimization", "author": "Fahad Sherwani", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Universiti Tun Hussein Onn Malaysia", "university_source": "repository_host", "college": "", "department": "", "country": "Malaysia", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "medical-ai", "research_subdomain": "rnn", "categories": "machine-learning; deep-learning; medical-ai; optimization; rnn", "keywords": "", "abstract": "Particle Swarm Optimization (PSO) is a population based stochastic optimization technique which consist of particles that move collectively in iterations to search for the most optimum solutions. However, conventional PSO is prone to lack of convergence and even stagnation in complex high dimensional-search problems with multiple local optima. Therefore, this research proposed an improved Mutually-Optimized Fractional PSO (MOFPSO) algorithm based on fractional derivatives and small step lengths to ensure convergence to global optima by supplying a fine balance between exploration and exploitation. The proposed algorithm is tested and verified for optimization performance comparison on ten benchmark functions against six existing established algorithms in terms of Mean of Error and Standard Deviation values. The proposed MOFPSO algorithm demonstrated lowest Mean of Error values during the optimization on all benchmark functions through all 30 runs (Ackley = 0.2, Rosenbrock = 0.2, Bohachevsky = 9.36E-06, Easom = -0.95, Griewank = 0.01, Rastrigin = 2.5E-03, Schaffer = 1.31E-06, Schwefel 1.2 = 3.2E-05, Sphere = 8.36E-03, Step = 0). Furthermore, the proposed MOFPSO algorithm is hybridized with Back-Propagation (BP), Elman Recurrent Neural Networks (RNN) and Levenberg-Marquardt (LM) Artificial Neural Networks (ANNs) to propose an enhanced data classification framework, especially for data classification applications. The proposed classification framework is then evaluated for classification accuracy, computational time and Mean Squared Error on five benchmark datasets against seven existing techniques. It can be concluded from the simulation results that the proposed MOFPSO-ERNN classification algorithm demonstrated good classification performance in terms of classification accuracy (Breast Cancer = 99.01%, EEG = 99.99%, PIMA Indian Diabetes = 99.37%, Iris = 99.6%, Thyroid = 99.88%) as compared to the existing hybrid classification techniques. Hence, the proposed technique can be employed to improve the overall classification accuracy and reduce the computational time in data classification applications.", "source_url": "https:\/\/openalex.org\/W3113818546", "pdf_url": "http:\/\/eprints.uthm.edu.my\/61\/1\/24p%20FAHAD%20SHERWANI.pdf", "doi": "", "handle": "", "repository_name": "UTHM Institutional Repository (Universiti Tun Hussein Onn Malaysia)", "source_record_id": "W3113818546", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 824, "title": "Implementation of a surrogate model for structural optimization based on artificial neural network and genetic algorithm", "title_ar": "", "title_en": "Implementation of a surrogate model for structural optimization based on artificial neural network and genetic algorithm", "author": "Iuri Hermes Müller", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "deep-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; deep-learning; robotics; ai-applications; predictive-analytics; optimization; ai-industry", "keywords": "", "abstract": "The use of computational optimization methods in structural design, employing surrogate models based on artificial neural networks (ANNs), is a promising approach for achieving efficient and cost-effective solutions compared to traditional optimization relying solely on the finite element method (FEM). These methods are particularly effective for composite material structures, which can be tailored to meet varying mechanical requirements and are fabricated via fused deposition modeling (FDM), such as 3D printing, due to their flexibility and capability to efficiently produce complex geometries. This study aims to implement a numerical optimization algorithm combined with an ANN trained and tested using a database to optimize the support arm of an unmanned aerial vehicle (UAV), manufactured using FDM. The methodology was divided into two main stages. In the first stage, an algorithm was developed in Python to generate a database from FEM simulations conducted in ABAQUS software. These simulations considered the arm, fabricated using polylactic acid (PLA) as the constituent material, with one fixed end and the other subjected to thrust and torsional loads, replicating the UAV's operational conditions. Four geometric parameters of the arm (structure height and width, thickness, and the number of diagonal reinforcement cells) were modified to obtain the masses and stiffness values for each configuration. In the second stage, the algorithm was enhanced by incorporating three additional geometric variables (thickness of the edges and bottom reinforcement, and width of the top reinforcement), resulting in a more extensive database. This expanded dataset was used to train the ANN, which predicted masses and stiffness values based on the geometric variables. The multi-objective optimization problem was then defined using the Non-Dominated Sorting Genetic Algorithm II (NSGA-II) to minimize mass and maximize stiffness, constrained by the geometric variables, with the ANN employed to evaluate the multi-objective function as a surrogate for FEM. From the solutions obtained on the Pareto front, one was selected for experimental validation. The optimized and standard models were designed in computer-aided design (CAD) software, fabricated using FDM with PLA, and subjected to bending tests to replicate the conditions implemented in the FEM analyses. The results demonstrated that the ANN exhibited high accuracy, with correlation coefficients of 0.99 for stiffness and 0.99 for mass, reducing analysis time by 99.65% compared to FEM. The optimized model showed an average specific stiffness increase of 51.14% and a mass reduction of 42.13% compared to the standard model. Finally, the results indicated that the proposed methodology is efficient in determining optimized solutions. While the database generation required substantial computational time, analyses conducted using the trained neural network demonstrated a significant reduction in processing time. Moreover, the proposed method is adaptable for application in various engineering projects involving composite materials", "source_url": "https:\/\/openalex.org\/W7120726491", "pdf_url": "http:\/\/repositorio.ufsm.br\/bitstream\/1\/33781\/1\/DIS_PPGEM_2024_MULLER_IURI.pdf", "doi": "", "handle": "", "repository_name": "LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas)", "source_record_id": "W7120726491", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 832, "title": "Weakly supervised learning for medical image segmentation", "title_ar": "", "title_en": "Weakly supervised learning for medical image segmentation", "author": "Hateef Alshewaier", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Cardiff University", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; responsible-ai; predictive-analytics; image-segmentation; medical-imaging; fairness", "keywords": "", "abstract": "Medical image segmentation plays a critical role in clinical diagnosis and treatment planning; however, the performance of deep learning models typically depends on large quantities of pixel-wise annotated data, which are costly and time-consuming to obtain. This thesis investigates annotation-efficient approaches for medical image segmentation by developing weakly supervised learning frameworks that reduce annotation requirements while maintaining high segmentation accuracy. The first contribution proposes an enhanced scribble-based segmentation framework that extends a modified U-Net architecture with increased feature learning capacity, mix-based data augmentation, and global and local consistency regularisation. These components improve learning from sparse scribble annotations and promote anatomically coherent segmentation masks. Experimental results demonstrate improved segmentation performance compared with existing scribble-supervised methods. The second contribution presents an ensemble learning framework for weakly supervised medical image segmentation based on single bounding-box annotations. The proposed framework integrates U-Net, ResUNet, and DeepLabV3 using bagging and boosting strategies to exploit complementary architectural characteristics. Despite relying solely on coarse single-bounding-box supervision, the ensemble models improve robustness, reduce model-specific bias, and yield more stable segmentation predictions. Experimental results demonstrate that the ensemble methods consistently outperform the individual models, with boosting achieving the highest segmentation performance. The third contribution introduces a novel dual bounding-box supervision strategy that utilises an inner bounding box to represent high-confidence object regions and an outer bounding box to provide contextual anatomical information. A region-aware weighted loss function encourages accurate object localisation while reducing ambiguity in boundary regions. The proposed framework significantly outperforms conventional single-box weakly supervised approaches and narrows the gap between weak and full supervision. Overall, the findings demonstrate that high-quality medical image segmentation can be achieved using sparse scribble annotations, ensemble learning techniques, and coarse bounding-box supervision without requiring exhaustive pixel-wise labels. The proposed methods reduce annotation effort while maintaining competitive segmentation accuracy, providing a practical and scalable pathway for deploying deep learning-based segmentation systems in clinical environments.", "source_url": "https:\/\/openalex.org\/W7219256121", "pdf_url": "", "doi": "", "handle": "", "repository_name": "ORCA Online Research @Cardiff (Cardiff University)", "source_record_id": "W7219256121", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 833, "title": "Deep Learning and Scientific Machine Learning For Medical Imaging Applications", "title_ar": "", "title_en": "Deep Learning and Scientific Machine Learning For Medical Imaging Applications", "author": "Adrian Celaya", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Rice University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; predictive-analytics; cnn; gnn; image-segmentation; medical-imaging", "keywords": "", "abstract": "Medical imaging technologies, such as magnetic resonance imaging (MRI) and computed tomography (CT), are invaluable tools in modern medicine - allowing for safe, non-invasive diagnosis and treatment planning. However, leveraging these imaging techniques for scientific modeling often involves a complex process that includes image acquisition, labeling (i.e., segmenting) regions of interest, generating a computational mesh, and numerical simulation. The contributions of this thesis focus on two areas within this pipeline: deep learning-based methods for efficient, scalable, and accurate 3D medical image segmentation and scientific machine learning (SciML) for solving partial differential equations (PDEs). In the realm of medical imaging segmentation, this thesis presents two innovative architectures that are based on geometric multigrid methods: the FMG and W-Nets. These novel architectures improve the segmentation accuracy compared to current state-of-the-art architectures with fewer parameters and converge to lower loss function values in fewer iterations. This thesis goes on to present the Generalized Surface Loss function, a novel, boundary-aware loss function that improves the accuracy of deep learning-based medical imaging segmentation. These innovations are incorporated into a scalable, open-source framework called the Medical Imaging Segmentation Toolkit, created to empower researchers to develop and test new concepts, such as novel architectures and loss functions, in a simple and reproducible pipeline. The presentation of work related to SciML begins by introducing a novel adaptive collocation point selection strategy for Physics-Informed Neural Networks (PINNs) using the QR Discrete Empirical Interpolation Method (QR-DEIM). This method improves both the training efficiency (i.e., convergence to lower loss values in fewer iterations) and predictive accuracy of PINNs by enabling more informed and targeted sampling of collocation points, particularly in regions of high solution complexity. Additionally, this work demonstrates how convolutional neural networks (CNNs) can approximate finite difference and discontinuous Galerkin (DG) solutions to elliptic and parabolic PDEs, both in supervised and unsupervised settings. These results highlight how SciML techniques can enhance traditional numerical solvers. The thesis culminates in the development of graph neural networks (GNNs) for approximating DG solutions over general domains. By representing computational meshes as graphs, these GNNs can generalize across varying resolutions and topologies, offering a flexible tool for scientific computing in irregular domains. Together, the contributions in this dissertation bridge the gap between deep learning, scientific machine learning, and classical numerical methods, offering a cohesive framework for deploying deep learning and SciML in medical imaging and broader computational physics applications.", "source_url": "https:\/\/openalex.org\/W7213488376", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Rice Research Repository (Rice University)", "source_record_id": "W7213488376", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 836, "title": "Deep Learning in Medical Image Analysis: Alzheimer’s Disease Classification & Whole Fetal Brain Segmentation.", "title_ar": "", "title_en": "Deep Learning in Medical Image Analysis: Alzheimer’s Disease Classification & Whole Fetal Brain Segmentation.", "author": "Jesús Alejandro Valdés Valdés", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Tecnológico de Monterrey", "university_source": "repository_host", "college": "", "department": "", "country": "Mexico", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; generative-ai; medical-ai; transfer-learning; cnn; image-segmentation; gan; medical-imaging", "keywords": "", "abstract": "During the last couple years there has been a growing interest in Deep Learning (DL), thanks in part to an increasing availability of computational power and of massive labeled datasets. This work presents two different Deep Learning applications for medical image analysis related tasks. The first application is the classification of Alzheimer Disease (AD) using MRI. AD is the most common form of dementia. However, it’s difficult to diagnose. During this work three different DL models are proposed to classify subjects into either AD or Normal Control (NC) groups. The models proposed are a custom CNN model, vgg16 + Transfer Learning and vgg16 + Fine Tuning. The vgg16 based methods make use of pre-learned weights trained on the ImageNet dataset. All of the models used as an input a single 2D MRI slice. This helped reduced model complexity and training times. After performing a 10 times repeated 5-Fold cross validation it was discovered that the best model was vgg16 with Fine Tuning. It reached an accuracy of 80% and an AUC of 0.86. The obtained results proved that the use of weights learned from different domains could be useful in medical image applications. The main contributions of this section were the use of a single MRI slice to classify patients, the model validation technique and the use of transfer learning. The second application is the fully automatization of the Whole Fetal Brain Segmentation in maternal MRI scans. For this U-Net [41] and other segmentation networks were compared. The U-Net model was also modified to include two types of attention components. Squeeze and Excitation module [16] and Attention Gates [39]. To avoid overfitting the models were trained using rigorous regularization (channel wise dropout, weight decay and data augmentation). They were also trained using different loss functions, including a Hybrid (BCE+Dice) loss function. The results of the 10-fold cross validation experiments showed statistically that no model was able to outperform the vanilla U-Net (DSC of 0.93 and HD 7.59mm). The proposed Attention Gated U-Net trained with the Hybrid loss reached a 0.95 DSC and 5.03mm HD (no significant difference). However, results on challenging data show a better performance for the Attention Gated models. This led to the conclusion that the possible advantage of using attention gates outweighed the cost of adding the attention module. The results were also improved by the use of a post processing step based on traditional computer vision techniques. The main contributions to this problem were the use of 10-fold cross validation, the statistical comparison between models, the use of the Hausdorff distance metric, the post processing step and the addition of attention components. This work also provides an overview of other deep learning applications also used within a medical setting. The techniques mentioned include image labeling, autoencoders and GANs for data augmentation. It also provides an overview of the current challenges and limitations DL faces and the most common way of mitigating them. Thanks to the research conducted during this thesis it can be concluded that the use of DL can be used in a medical setting focused on the analysis of images. Particularly on classification and segmentation tasks. However, certain challenges such as having a large enough data set and availability of the required hardware to carry out research must be considered before attempting to solve a problem with DL.", "source_url": "https:\/\/openalex.org\/W3014475976", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Repositorio del Tecnológico de Monterrey (Tecnológico de Monterrey)", "source_record_id": "W3014475976", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 840, "title": "Anatomy-aided deep learning for Medical Image Segmentation : from the anatomy of the epicardial adipose tissue", "title_ar": "", "title_en": "Anatomy-aided deep learning for Medical Image Segmentation : from the anatomy of the epicardial adipose tissue", "author": "Lu Liu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; medical-ai; image-segmentation; medical-imaging", "keywords": "", "abstract": "Medical image segmentation plays a pivotal role in modern healthcare by enabling precise diagnosis, treatment planning, and continuous monitoring of diseases. Among various segmentation tasks, the segmentation of epicardial adipose tissue (EAT) in low-dose computed tomography (LDCT) scans has gained attention due to its relevance in cardiovascular health. EAT is a vital fat depot surrounding the heart, closely associated with inflammatory mediators and cytokines, making it a crucial imaging biomarker for various cardiovascular conditions. LDCT is particularly suited for EAT segmentation, offering detailed anatomical information with minimal radiation exposure, thus allowing for repeated scans and large-scale studies. These studies are essential for assessing cardiovascular risk factors, including EAT volume, and identifying multiple biomarkers in a single scan for comprehensive cardiovascular health evaluations. However, segmenting EAT in LDCT scans presents significant challenges. The irregular shape and non-uniform distribution of EAT around the heart make it difficult to distinguish from surrounding tissues. Additionally, LDCT imaging is prone to noise and artifacts, which obscure image clarity and complicate the accurate delineation of EAT boundaries. The lack of specialized annotated datasets further hampers the development of reliable segmentation algorithms. Although advanced deep learning approaches have been applied to EAT segmentation, they often require large annotated datasets, are sensitive to noise, and struggle with interpreting features without explicit anatomical priors. To address these challenges, this thesis emphasizes the integration of anatomical information, particularly the shape of the pericardium, into segmentation methodologies. This approach aims to improve the accuracy, efficiency, and clinical relevance of EAT segmentation, ultimately contributing to better patient care and more informed medical decision-making. The thesis is built on a foundation of thorough literature review, examining existing anatomy-aided deep learning techniques for medical image segmentation. A significant contribution of the research is the creation of a specialized dataset consisting of 154 LDCT scans, annotated with pericardium and pixel-wise EAT labels. This dataset serves as a critical resource for training and evaluating segmentation algorithms. The thesis explores various deep learning models, such as 3D U-net and its variants, and evaluates their performance using traditional metrics like the Dice coefficient and statistical analyses like Pearson correlation and Bland-Altman analysis. A novel shape-based loss function derived from Fourier descriptors is introduced to enhance segmentation accuracy, particularly in challenging scenarios. Additionally, the thesis explores the use of Fourier neural operators to generalize shape encoding in the Fourier space, addressing the limitations of traditional methods and improving the generalization capabilities of models across different medical image segmentation tasks. Through these advancements, the thesis aims to provide state-of-the-art tools for healthcare professionals, driving progress in medical image segmentation and contributing to enhanced clinical outcomes.", "source_url": "https:\/\/openalex.org\/W4400468663", "pdf_url": "", "doi": "10.3990\/1.9789036562072", "handle": "", "repository_name": "", "source_record_id": "W4400468663", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 842, "title": "Robust liver vessel segmentation in medical images using 3-D deep learning approaches", "title_ar": "", "title_en": "", "author": "Abir Affane", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2022, "language": "fr", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; medical-ai; image-segmentation; medical-imaging", "keywords": "", "abstract": "Une approche Deep Learning pour la segmentation robuste des vaisseaux hépatiques sur des images médicales 3-D L’évolution de l’informatique dans la recherche sur les réseaux vasculaires a permis d’améliorer la reconstruction et l’interprétation de ces structures complexes.L’extraction robuste de ces structures sur des images biomédicales nécessite une bonne acquisition lors d’un examen d’imagerie.La segmentation des vaisseaux peut être utilisée pour de nombreuses applications telles que la division du foie en sous-régions. Cette définition des segments de Couinaud a été une bonne initialisation pour les chercheurs pour résoudre plu-sieurs problèmes liés aux maladies du foie, faciliter l’interprétation des images parles radiologues et même pour des planifications chirurgicales. L’exploration des vaisseaux hépatiques repose sur la bonne acquisition lors d’un examen. Le deep learning apporte de nouvelles solutions aux problèmes de segmentation des images médicales et devient une méthodologie de pointe pour de futures applications cliniques. L’étude des structures vasculaires est une tâche difficile en raison de la taille extrêmement petite de la structure du vaisseau, du faible SNR (Signal to Noise Ratio) et du contraste variable des données d’imagerie médicale. L’étude des structures curvilignes est une tâche difficile, en particulier compte tenu des applications d’imagerie médicale telles que la reconstruction des vaisseaux. Depuis l’arrivée des méthodes d’apprentissage profond, plusieurs approches basées sur l’intensité des pixels ont été proposées pour réduire les erreurs de segmentation, mais ces informations ne sont toujours pas suffisantes pour préserver la connectivité au sein des branches et peuvent donner de mauvais résultats de segmentation dans les bifurcations. Ceci est principalement dû à la complexité de ces structures et au contraste inhomogène dans tout le réseau vasculaire des images médicales.Cette thèse est consacrée au développement d’une nouvelle méthode d’apprentissage en profondeur 3-D basée sur des informations topologiques pour la segmentation des vaisseaux hépatiques à partir des images médicales, qui peut aider à améliorer les décisions de segmentation à partir des modèles vasculaires. Ces approches intègrent à la fois l’intensité du pixel et la structure topologique des vaisseaux en développant des modèles 3-D avec une fonction de perte topologique spécifique, qui calcule la similarité entre la carte de vraisemblance prédite par un modèle 3-D de segmentation et les vaisseaux de vérité de terrain. Notre modèle d’apprentissage topologique profond 3-D est une technique prometteuse avec une meilleure qualité de segmentation sur les bifurcations notamment.Dans une première partie, des modèles de segmentation U-Net 3-D sont revisités et comparés. Plus précisément, 3-D U-Net, Dense U-Net et MultiRes U-Net sont opposés les uns aux autres dans la tâche de segmentation des vaisseaux sur le jeu de données scanner public de l’IRCAD. Pour chaque modèle, trois configurations alternatives permettant d’adapter les architectures du réseau neuronal convolutif adaptées aux données volumétriques sont testées. À savoir, les configurations 3-Dcomplètes, basées sur des groupes de coupes (slabs) et sur des patches 3-D sont prises en compte. Ces méthodes permettent de mieux détecter les bifurcations.Cette comparaison des modèles 3-D permet de choisir la meilleure approche basée sur les intensités de l’image.Dans une seconde partie, une méthode basée sur la combinaison des modèles d’apprentissage profond avec des filtres de rehaussement est introduite pour segmenter les vaisseaux hépatiques sur des images 3-D. Cette approche améliore les résultats de segmentation. Plusieurs filtres de rehaussement ont été utilisés et comparés. Enfin la troisième partie de la thèse est consacrée à introduire les signatures topo-logiques. Ces informations peuvent aider à améliorer la décision de segmentation.L’étude hiérarchique du squelette (sous forme de graphe) permet de donner une description graphique du réseau vasculaire hépatique. (...)", "source_url": "https:\/\/openalex.org\/W7151558747", "pdf_url": "", "doi": "10.70675\/e29c06baza80dz47ebza090z3fe9d7c42991", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7151558747", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 845, "title": "Exploring the Potential of Deep Learning for Biomedical Image Segmentation", "title_ar": "", "title_en": "Exploring the Potential of Deep Learning for Biomedical Image Segmentation", "author": "Sharif, Muhammad Hamza", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United Arab Emirates", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "deep-learning; computer-vision; medical-ai; cnn; transformers; image-segmentation; medical-imaging", "keywords": "", "abstract": "Medical image segmentation is the fundamental task in biomedical imaging analysis by identifying and isolating regions of interest from images such as MRI or CT scans. This process is vital for precise diagnosis and effective treatment planning for various conditions, including tumors and anomalies. However, medical image segmentation is a challenging task due to the complex and diverse nature of these images. They contain structures with varying sizes, shapes, and intensities, as well as noise, artifacts, and missing data. Traditional image segmentation methods that rely on handcrafted features and heuristic algorithms have limitations in accurately segmenting medical images due to their complexity and variability. Recent advances in deep learning-based techniques have shown promising results in medical image segmentation by automatically learning feature representations. However, these techniques require innovation in network architecture, learning framework, and training procedures to improve the segmentation process. In this thesis, we aim to introduce new deep learning-based methods to improve medical image segmentation, with a focus on frameworks for high-resolution and volumetric medical image segmentation. The proposed techniques aim to segment various structures, including organs, brain tumors, skin lesions, polyps, and retinal vessels, using medical image data from different modalities such as CT, MRI, endoscopy imaging, and fundus imaging. The thesis proposes two novel deep learning-based techniques. The first one is called TransResNet, which combines a transformer and CNN to effectively segment high-resolution medical images. The proposed method introduces a Cross Grafting Module (CGM) that generates grafted features by combining global semantic and low-level spatial details. This approach achieves state-of-the-art or competitive results on several segmentation tasks, including skin lesions, retinal vessels, and polyp segmentation. The second technique proposed in the thesis is a versatile generic architecture called Y-CA-Net, which can segment volumetric medical images of different modalities (CT, MRI). This method is based on any two encoders and a decoder backbone, with one encoder branch that extracts local features using convolution filters and the other encoder branch that learns global interactions using an attention mechanism. The proposed approach aims to fully exploit the strengths of both convolution and attention mechanisms for volumetric segmentation. The model introduces a simple yet effective Cross Feature Mixer Module (CFMM) to learn semantically enriched features by mixing local and global feature representations. The proposed model has shown significant improvement across different benchmark datasets for multi-organ and brain tumor segmentation compared to existing state-of-the-art methods.", "source_url": "https:\/\/openalex.org\/W7106329223", "pdf_url": "", "doi": "10.82419\/162", "handle": "", "repository_name": "MBZUAI iRep", "source_record_id": "W7106329223", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 846, "title": "Learning with imperfect datasets in medical image segmentation", "title_ar": "", "title_en": "Learning with imperfect datasets in medical image segmentation", "author": "Yuhang Ding", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Technology Sydney", "university_source": "repository_host", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; responsible-ai; ai-applications; predictive-analytics; image-segmentation; medical-imaging; fairness; ai-industry", "keywords": "", "abstract": "edical image segmentation aims to partition medical images into distinct physiological regions, such as organs and lesions, which is crucial for disease diagnosis and treatment planning.The advent of deep neural networks has significantly advanced this field.However, the performance in real-world scenarios remains unsatisfactory due to imperfect data and annotations during both training and deployment.First, scaling up training data and annotations is challenging.This is because obtaining medical images is difficult due to privacy concerns.Furthermore, annotating medical images requires substantial expertise, making the process costly and difficult to carry out.Second, the quality of medical images cannot always be guaranteed in real-world scenarios, leading to significant performance drops in outlier cases.Third, real-world medical applications are safety-critical and demand extremely accurate predictions, a requirement that most existing models fail to meet adequately.These challenges hinder the practical deployment of medical image segmentation.Consequently, both academic and industrial communities are striving to develop highly accurate medical image segmentation algorithms that can perform well despite imperfect data.To this end, this thesis proposes deep learning methods to develop a well-performed medical image segmentation model that can be effectively trained with limited and low-quality data\/annotations.Specifically, a comprehensive suite is proposed from three directions: (1) applying image registration to generate realistic and diverse training samples and adopting barely-supervised learning paradigms to enable learning with insufficient annotated data; (2) devising region-aware fusion module to address missing modality problem; (3) incorporating automatic and interactive medical image segmentation into a single model and one training session to achieve sufficient segmentation performance for practical use.Extensive experiments on several medical image segmentation tasks, such as brain tumor segmentation, brain structure segmentation and abdominal organ segmentation, demonstrate the effectiveness and efficiency of the proposed techniques.ix", "source_url": "https:\/\/openalex.org\/W7216881435", "pdf_url": "https:\/\/opus.lib.uts.edu.au\/bitstream\/10453\/186717\/1\/thesis.pdf", "doi": "", "handle": "", "repository_name": "OPUS - Open Publications of UTS Scholars (University of Technology Sydney)", "source_record_id": "W7216881435", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 848, "title": "Constrained deep networks for medical image segmentation", "title_ar": "", "title_en": "Constrained deep networks for medical image segmentation", "author": "Hoel Kervadec", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; self-semi-supervised; predictive-analytics; cnn; image-segmentation; medical-imaging", "keywords": "", "abstract": "Weakly supervised image segmentation, in the form of partially labeled images, is attracting significant research attention as it can mitigate the need for laborious pixel annotations required by deep learning models. Enforcing high-order, global inequality constraints on the network outputs can leverage unlabeled data by guiding the training with prior knowledge, restricting the search space during training to anatomically feasible solutions. A range of possible values (such as a lower\/upper bounds on the size of a organ) can be very valuable to guide training. However, in the context of deep neural networks, standard Lagrangian optimization has been largely avoided, mainly due to the instability and computational complexity ensuing from alternating explicit dual updates and stochastic optimization. Interior point methods, despite their popularity in convex optimization, are not applicable neither, as they require a feasible starting point, which is itself a difficult constrained problem for deep neural networks. In this thesis, we investigate hard inequality constraints in the context of deep networks with both quadratic penalties and more principled log-barrier extensions. We also investigate methods to mitigate class-imbalance in segmentation problems, such as in brain lesions dataset, by constraining the boundary of the predicted segmentation to match the ground-truth boundary. This thesis produced five different publications as first author, and four papers as co-author. Our papers received several awards, and we were invited to publish extended versions of our works in two special issues of Medical Image Analysis (MedIA). In our first contribution, we propose to introduce a differentiable penalty, which enforces inequality constraints directly in the loss function, avoiding expensive Lagrangian dual iterates and proposal generation. From constrained-optimization perspective, our simple penalty-based approach is not optimal as there is no guarantee that the constraints are satisfied. However, surprisingly, it yields substantially better results than Lagrangian-based constrained convolutional neural networks, while reducing the computational demand for training. By annotating only a small fraction of the pixels, our approach reaches performances comparable to full supervision, on three separate tasks. While our experiments focused on basic linear constraints such as the target-region size and image tags, our framework can be easily extended to other non-linear constraints, e.g., invariant shape moments and other region statistics. In our second contribution, we propose log-barrier extensions, which approximate Lagrangian optimization of constrained-CNN problems with a sequence of unconstrained losses. Unlike standard interior-point and log-barrier methods, our formulation does not need an initial feasible solution. We report comprehensive weakly supervised segmentation experiments, with various constraints, showing that our formulation outperforms substantially the existing constrained-CNN methods, both in terms of accuracy, constraint satisfaction and training stability. In our third contribution, we enforce constraints on the boundary of predicted segmentation. Widely used loss functions for CNN segmentation, such as Dice or cross entropy, are based on integrals over the segmentation regions. Unfortunately, for highly unbalanced segmentations, such regional summations have values that differ by several orders of magnitude across classes, which affects training performance and stability. We propose a boundary loss, which takes the form of a distance metric on the space of contours, not regions. This can mitigate the difficulties of highly unbalanced problems because it uses integrals over the interface between regions instead of unbalanced integrals over the regions. Furthermore, a boundary loss complements regional information. Inspired by graph-based optimization techniques for computing activecontour flows, we express a non-symmetric L2 distance on the space of contours as a regional integral, which avoids completely local differential computations involving contour points. This yields a boundary loss expressed with the regional softmax probability outputs of the network, which can be easily combined with standard regional losses and implemented with any existing deep network architecture for N-D segmentation. We report comprehensive evaluations on different unbalanced problems, showing that our boundary loss can yield significant increases in performances while improving training stability. In a fourth contribution, we investigates a curriculum-style strategy for semi-supervised CNN segmentation, which devises a regression network to learn image-level information such as the size of the target region. These regressions are used to effectively regularize the segmentation network, constraining the softmax predictions of the unlabeled images to match the inferred label distributions.", "source_url": "https:\/\/openalex.org\/W3215133863", "pdf_url": "https:\/\/espace.etsmtl.ca\/id\/eprint\/2734\/1\/KERVADEC_Hoel.pdf", "doi": "", "handle": "", "repository_name": "Espace ÉTS (ETS)", "source_record_id": "W3215133863", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 851, "title": "Deep Learning for Pneumothorax Segmentation and Quantification", "title_ar": "", "title_en": "Deep Learning for Pneumothorax Segmentation and Quantification", "author": "Wannipa Sae-Lim", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Prince of Songkla University", "university_source": "repository_host", "college": "", "department": "", "country": "Thailand", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; transfer-learning; predictive-analytics; cnn; transformers; image-segmentation; medical-imaging", "keywords": "", "abstract": "This thesis investigates the role of deep learning, focusing on convolutional neural networks and encoder-decoder architectures, to enhance medical image segmentation, specifically for pneumothorax identification in chest X-ray images. Pneumothorax, a critical condition characterized by lung collapse, requires precise diagnostics for effective management and treatment. Traditional image segmentation methods sometimes underperform due to the complexity and variability of chest X-ray images. In this study, we proposed PTXSeg-Net, a novel deep learning model for accurate and efficient pneumothorax segmentation. This model incorporates residual blocks and attention mechanisms to improve learning capabilities and is further enhanced by deep supervision, which allows for more effective gradient utilization across all network layers, significantly enhancing its performance over existing segmentation methods. Additionally, transfer learning with pre-trained models on extensive chest X-ray datasets improves feature extraction, addressing imbalanced datasets. Data refinement techniques are also employed to optimize the training outcomes. The performance of PTXSeg-Net outperforms the existing segmentation model in terms of accuracy, dice and Jaccard coefficients, demonstrating significant improvement in the field of pneumothorax segmentation. Moreover, leveraging the predicted masks of pneumothorax, this thesis proposes an algorithm that calculates the pneumothorax ratio. This algorithm provides a quantitative measure of the appearance of pneumothorax relative to the total lung area, offering a metric for radiologists to assess the severity of the condition more accurately. This integration of deep learning models and practical quantification algorithm underscores the significant contribution to the field of medical image analysis.", "source_url": "https:\/\/openalex.org\/W7201935903", "pdf_url": "", "doi": "", "handle": "", "repository_name": "PSU Knowledge Bank (Prince of Songkla University)", "source_record_id": "W7201935903", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 853, "title": "A hybrid Convolutional Neural Network (CNN) and Support Vector Machine (SVM) for detecting Keratoconus in patients using genetic algorithm", "title_ar": "", "title_en": "A hybrid Convolutional Neural Network (CNN) and Support Vector Machine (SVM) for detecting Keratoconus in patients using genetic algorithm", "author": "Somayeh Ghasedi Katak Lahijani", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Laurentian University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "cnn", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; predictive-analytics; optimization; cnn", "keywords": "", "abstract": "Nowadays, there are still significant challenges encountered in the accurate diagnosis of various eye diseases, such as Keratoconus (KCN) and cataracts. Early detection of Keratoconus is crucial in preventing its progression and ensuring the best treatment outcomes. Artificial intelligence (AI) is being widely applied in ophthalmology through the training of deep learning networks for the early detection of eye diseases. This research presents a novel, integrated machine learning approach for diagnosing Keratoconus disease by combining feature extraction through Convolutional Neural Networks (CNN) with a Support Vector Machine (SVM) and Artificial Neural Network (ANN) for classification. Employing a multi-objective genetic algorithm, the method optimizes feature selection, aiming to minimize both diagnostic error and the number of features. The study utilizes a dataset of 5,152 ophthalmic images (1288 samples) categorized into Normal (476), Suspect (453), and Keratoconus (359) cases, sourced from an eye clinic in Egypt. Combining a Convolutional Neural Network (CNN) for feature selection with a Genetic Algorithm (GA) significantly improved diagnostic accuracy. Consequently, by focusing on the most relevant features of Keratoconus (KC), the model achieved an impressive 98.63% accuracy for ANN classification with a genetic algorithm, and 98.13% for SVM classification with a genetic algorithm. The accuracy of the algorithm exceeded that of when SVM and ANN were used without the genetic algorithm alone, which were 97.53% and 96.9% respectively, underscoring the benefit of combining Artificial Neural Networks (ANNs) with Genetic Algorithms (GAs) in KC diagnosis. Implementing this model can assist physicians in more accurate Keratoconus detection, providing better predictions regarding patients' eye conditions, and offering timely treatment recommendations.", "source_url": "https:\/\/openalex.org\/W7205487330", "pdf_url": "https:\/\/laurentian.scholaris.ca\/bitstreams\/ebade523-4710-440b-a199-72030adbc718\/download", "doi": "", "handle": "", "repository_name": "Lu Zone Ul (Laurentian University)", "source_record_id": "W7205487330", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 856, "title": "Development of Automated Extraction Method for Road and Centerline from Aerial Images Using Deep Convolutional Neural Network", "title_ar": "", "title_en": "Development of Automated Extraction Method for Road and Centerline from Aerial Images Using Deep Convolutional Neural Network", "author": "Tamara Alshaikhli", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Japan", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "computer-vision", "research_subdomain": "image-segmentation", "categories": "deep-learning; computer-vision; ai-applications; cnn; image-segmentation; ai-smart-cities", "keywords": "", "abstract": "Road extraction from remote sensing imagery plays an important role in modern life society.That contributes in a lot of areas, such as urban planning, disaster mitigation, transportation engineering, and so on.Manual updating was the common methodology to keep the road information up to date, however the hectic process to label these images manually requires a lot of time and efforts, and is also prone to human error.Therefore, this area attracts a lot of researchers' attentions to find a suitable automated process.This research adopted a deep learning technique as an alternative to solve the problem of road extraction from aerial imagery automatically.The research consisted of two parts.The first one was dedicated to extract road only by a deep convolutional neural network (DCNN).The proposed DCNN was built on the U-Net backbone with a residual block (Res-block) on the encoder part and convolution layers (Conv.layer) on the decoder part.The model shows a good result for extracting road automatically from aerial images, and the research has extended to conduct a comprehensive study covering all the important aspects of this area.The second part of this research was assigned to extract road and centerline simultaneously by extending the previous model with multi task learning scheme (MTL).The proposed model is a complete MTL scheme without any further utilization of thinning algorithms or post processing.The proposed model showed a promising results for the single task learning (STL) to extract road only.In addition, the MTL for extracting road and centerline simultaneously also succeeded in extracting the roads and centerlines simultaneously.In order to evaluate the applicability of the proposed model for another dataset, a case study for Baghdad City, Iraq was performed, and the accuracy was examined.", "source_url": "https:\/\/openalex.org\/W7143882449", "pdf_url": "https:\/\/opac.ll.chiba-u.jp\/da\/curator\/900120235\/WLA_0058.pdf", "doi": "10.20776\/900120235", "handle": "", "repository_name": "Institutional Repositories DataBase (IRDB)", "source_record_id": "W7143882449", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 858, "title": "Flexible network management in software defined wireless sensor networks for monitoring application systems", "title_ar": "", "title_en": "Flexible network management in software defined wireless sensor networks for monitoring application systems", "author": "Kgotlaetsile Mathews Modieginyane", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Pretoria", "university_source": "repository_host", "college": "", "department": "", "country": "South Africa", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "computer-vision", "research_subdomain": "", "categories": "machine-learning; computer-vision", "keywords": "", "abstract": "Wireless Sensor Networks (WSNs) are the commonly applied information technologies of modern networking and computing platforms for application-specific systems. Today’s network computing applications are faced with high demand of reliable and powerful network functionalities. Hence, efficient network performance is central to the entire ecosystem, more especially where human life is a concern. However, effective management of WSNs remains a challenge due to problems supplemental to them. As a result, WSNs application systems such as in monitored environments, surveillance, aeronautics, medicine, processing and control, tend to suffer in terms of capacity to support compute intensive services due to limitations experienced on them. A recent technology shift proposes Software Defined Networking (SDN) for improving computing networks as well as enhancing network resource management, especially for life guarding systems. As an optimization strategy, a software-oriented approach for WSNs, known as Software Defined Wireless Sensor Network (SDWSN) is implemented to evolve, enhance and provide computing capacity to these resource constrained technologies. Software developmental strategies are applied with the focus to ensure efficient network management, introduce network flexibility and advance network innovation towards the maximum operation potential for WSNs application systems. The need to develop WSNs application systems which are powerful and scalable has grown tremendously due to their simplicity in implementation and application. Their nature of design serves as a potential direction for the much anticipated and resource abundant IoT networks. Information systems such as data analytics, shared computing resources, control systems, big data support, visualizations, system audits, artificial intelligence (AI), etc. are a necessity to everyday life of consumers. Such systems can greatly benefit from the SDN programmability strategy, in terms of improving how data is mined, analysed and committed to other parts of the system for greater functionality. This work proposes and implements SDN strategies for enhancing WSNs application systems especially for life critical systems. It also highlights implementation considerations for designing powerful WSNs application systems by focusing on system critical aspects that should not be disregarded when planning to improve core network functionalities. Due to their inherent challenges, WSN application systems lack robustness, reliability and scalability to support high computing demands. Anticipated systems must have greater capabilities to ubiquitously support many applications with flexible resources that can be easily accessed. To achieve this, such systems must incorporate powerful strategies for efficient data aggregation, query computations, communication and information presentation. The notion of applying machine learning methods to WSN systems is fairly new, though carries the potential to enhance WSN application technologies. This technological direction seeks to bring intelligent functionalities to WSN systems given the characteristics of wireless sensor nodes in terms of cooperative data transmission. With these technological aspects, a technical study is therefore conducted with a focus on WSN application systems as to how SDN strategies coupled with machine learning methods, can contribute with viable solutions on monitoring application systems to support and provide various applications and services with greater performance. To realize this, this work further proposes and implements machine learning (ML) methods coupled with SDN strategies to; enhance sensor data aggregation, introduce network flexibility, improve resource management, query processing and sensor information presentation. Hence, this work directly contributes to SDWSN strategies for monitoring application systems.", "source_url": "https:\/\/openalex.org\/W2949713261", "pdf_url": "https:\/\/repository.up.ac.za\/bitstreams\/ea51dda3-daff-4fe2-9a46-130b20288684\/download", "doi": "", "handle": "", "repository_name": "UpSpace Institutional Repository (University of Pretoria)", "source_record_id": "W2949713261", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 859, "title": "Üzüm salkımı meyvelerinin tanınması amacıyla DVM, ESA ve YSA sınıflayıcılarının gerçekleştirilmesi ve başarılarının belirlenmesi.", "title_ar": "", "title_en": "", "author": "Bashar Saad Falıh Al-Saffar", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Turkey", "governorate": "", "publication_year": 2018, "language": "tr", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; deep-learning; computer-vision; robotics; ai-applications; cnn; object-detection; ai-agriculture", "keywords": "", "abstract": "Grape detection, harvesting, spraying, and yield estimation are difficult, time consuming, and requiring considerable manpower.Therefore, harvesting with the aid of smart robots is now being explored and is currently a frontline problem in agriculture.This thesis presented a grape bunch and berry recognition by employing Convolution Neural Network (CNN), Artificial Neural Network (ANN) and Linear Support-Vector-Machine (SVM).Histogram of Oriented Gradients (HOG), the Local Binary Pattern (LBP), and combination have been used as feature vectors.The data contain berry and non-berry classes.ANN classifier with HOG+LBP features show better results in single berry recognition with an average of accuracy, precision and recall are 99.64±0.073%, 99.70±0.094%,and 99.59±0.117%respectively.The proposed detection method uses a Fast-Radial Symmetry Transform (FRST) as interest point detector.Feature extraction and classification are computed in each interest point approach.Afterwards, a DBSCAN method defines the number of grape bunches.Each cluster's spatial distribution and needs to close bunch separation for an improved bunch detection.The average accuracy of grape bunch detection 91.72±2.88% and 89.82±2.58%for berry detection.Results show a good performance based on the shape and texture information, with no need for special illumination and color features.", "source_url": "https:\/\/openalex.org\/W7132192735", "pdf_url": "http:\/\/library.cu.edu.tr\/tezler\/12625.pdf", "doi": "", "handle": "", "repository_name": "Çukurova University Institutional Repository", "source_record_id": "W7132192735", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 860, "title": "Modeling of Groundwater Heavy Metals and Methane Pollution for a Municipal Landfill Utilizing Data Driven and Numerical Modeling Techniques", "title_ar": "", "title_en": "Modeling of Groundwater Heavy Metals and Methane Pollution for a Municipal Landfill Utilizing Data Driven and Numerical Modeling Techniques", "author": "Bahareh Fallah", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "deep-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; deep-learning; predictive-analytics", "keywords": "", "abstract": "This thesis shows original results to the advancement of modeling of groundwater heavy metal pollution and methane generation as the main two concerns in landfilling for a municipal landfill. To better understand the context of thesis, it has been divided into six major studies. The first part of the thesis is on spatial-temporal analysis of dissolved metal pollutants near an unlined municipal landfill in a semi-arid climate. The data was collected from 5 groups of wells located upstream and downstream of the landfill in 2012 and 2015. The results represented that in 2015, Mn, U, As and Fe exceeded the Saskatchewan drinking water quality standard limits. However, in 2012 other than the aforementioned heavy metals, Cd concentration was also above the guideline. Mn concentration increased from 2012 to 2015 while As and Fe decreased in the same period. Correlation analysis was assessed among heavy metals and two linear regression models are developed for pH and dissolved Mn in groundwater. Field data and literature review suggested that groundwater contamination is highly site specific. In the second part of the thesis, using geostatistical analysis, the three water quality index-based maps were compared to a GWQ map from overlapping heavy metal kriged maps. All index and GWQ maps showed an elevated plateau extending from the unlined landfill to the southeast corner implying that the landfill may have negatively impacted groundwater quality. Using the index method, some wells have poor water quality with regards to Mn, especially at southeast of the landfill, and according to the GWQ map, the center part of the landfill exhibits high pollution in terms of As, that can be explained by correlation matrix and weighting method. The third part of the thesis focuses on prediction of organic waste generation, as the main source of methane emissions from municipal landfills, utilizing artificial neural networks. Significant socio-economic and climatic variables were used to develop organic waste generation models with different ratios of training and testing. Among the total of 30 ANN models, the best model used wheat, maximum humidity, and Nasdaq variables with train-test ratio of 90:10. The optimal numbers of neurons was 8 in the hidden layer with MAPE of 27.63% at the testing stage. In the fourth part of the thesis, the two-stage NARX neural network model with meteorological input parameters was developed to predict daily CH4 generation rate from the landfill. The effects of each pre-processing step including minimizing collinearity problems, data normalization and filtering the outliers was examined on improving the model performance. The model with MAPE of 3.03% at the testing stage, 3 layers, 5 climatic variables and 9 neurons in the hidden layer was the optimal model. Missing data imputation coupled with NARX neural network method decreased the MSE of the model by 84.3% (compared to MLP model) in the testing period representing the effectiveness of the proposed method in daily CH4 generation prediction. In the fifth part of the thesis, after data pre-processing, the daily ANN methane prediction model was compared to the LandGEM first order decay model with optimal parameters. The MSE of the MLP methane predictionmodel decreased by 74.642% by comparison to the LandGEM. The aggregated daily predicted data in the MLP model were more correlated to the yearly measured CH4 rate with R2 of 0.929 than that in the LandGEM model (R2 = 0.887). The LandGEM model even with the site-specific parameters significantly underestimated the measured CH4 rate. The sixth part of the thesis focuses on the most efficient length of training dataset and periodic-based methane prediction modeling. Using a stepwise procedure, various volumes of training data were assessed, and concluded that employing the three-year training data reduced the MSE of the testing stage by 51.29%. Using the single input DOY-based model with MAPE of 2.12% shows a higher performance with the MSE of 5.54% lower than that of the model with both climatic and periodic inputs showing a convenient and precise approach in landfill gas modeling because of its simplicity and high accuracy.", "source_url": "https:\/\/openalex.org\/W3204603059", "pdf_url": "https:\/\/ourspace.uregina.ca\/bitstream\/handle\/10294\/14329\/ShahidpourFallah_Bahareh_PhD_EVSE_Spring2021.pdf", "doi": "10.82465\/4991", "handle": "", "repository_name": "", "source_record_id": "W3204603059", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 861, "title": "The Role of Accumulated Dust Chemical Composition on Photovoltaic Thermal Performance: Advanced Modeling and Experimental Analysis", "title_ar": "", "title_en": "The Role of Accumulated Dust Chemical Composition on Photovoltaic Thermal Performance: Advanced Modeling and Experimental Analysis", "author": "Hussam M. Almukhtar", "supervisor": "", "degree_type": "phd", "degree_evidence": "نص السجل: «PhD»", "university": "Auckland University of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "New Zealand", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "deep-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; deep-learning; ai-applications; predictive-analytics; ai-smart-cities", "keywords": "", "abstract": "One of the significant challenges affecting photovoltaic (PV) performance is the influence of environmental conditions. Cloud cover, rain, and high temperatures can reduce PV efficiency. Among these, dust impacts PV performance and leads to secondary effects, such as increased PV temperature. This PhD thesis delves deeply into the impact of dust accumulation on photovoltaic (PV) modules, focusing on their thermal and optical performances. Firstly, the study identifies gaps in existing research and highlights the lack of consideration of how dust accumulation can impact PV temperature and the mathematical and specific correlation between dust accumulated weight and type on PV temperature. Additionally, the literature review reveals that a generalized module calculation is required to determine dust optical properties, which can help researchers and scholars identify the impact of dust on PV optical properties. This study introduces a comprehensive approach that integrates a wide range of dust characteristics into the thermal and optical analysis of PV modules. This research unfolds through several phases, each delving into distinct yet interconnected aspects of how dust affects PV efficiency. Regarding thermal effects, the investigation exposes the limitations of existing methodologies inadequately accounting for dust parameters. This revelation propels the research forward, employing advanced experimental setups and innovative mathematical models. These models, distinguished by their inclusion of absorption, emittance, and meteorological data, surpass traditional approaches in forecasting PV temperatures across varied dust conditions. The meticulous empirical validation against real-world scenarios underscores these novel formulations' heightened accuracy and practicality. Tested under real weather conditions, these models demonstrate a notable improvement in predicting PV module temperatures, achieving the lowest mean absolute error (MAE) of 1.4 compared to traditional methods, marking a significant advancement in the field. In the subsequent phase, the research underscored the optical considerations that consider essential parameters to determine the PV temperature. The study has employed intelligent methodologies such as Artificial Neural Networks (ANN) and Multiple Linear Regression (MLR) to develop a novel model. Those models incorporate dust weight and the principal elements in the chemical composition of dust particles. This segment of the study, drawing from over 300 experimental observations and leveraging more than 600 global datasets, has yielded notable predictive precision which can predict the dusty PV surface the most important dust optical properties, which are emissivity, absorption, and transmittance where those parameters can play a crucial role to predict the PV performance and temperature. Additionally, the predictive novel has been compared with other modules that have been developed using different techniques to produce the transmittance of dusty PV using The Mean Squared Errors (MSEs). This developed module achieved the lowest MSE, where 1.8 is achieved for the ANN models and 8.44 for the MLR models, setting new benchmarks for predictive accuracy within dust's influence on PV modules. The culmination of experimental insights and enhanced predictive accuracy constitutes the foundation of this thesis, providing a comprehensive perspective on the impact of dust on PV modules. This study addresses several critical research gaps, including the lack of specific models that integrate both the thermal and optical effects of dust accumulation, the absence of clear correlations between dust properties and PV temperature, and the need for generalized methods to quantify dust's optical properties (such as transmittance, absorption, and emissivity). These contributions, supported by empirical validation and advanced predictive modelling, significantly advance understanding and mitigating dust's effects on PV systems. The research's original contributions are validated through comparative analysis with existing literature, showcasing its superior predictive capabilities. This thesis significantly contributes to the field of PV technology by addressing crucial knowledge gaps and introducing improved methodological frameworks. It underscores the significance of a thorough comprehension of dust's profound influence on PV modules, advocating for a multidisciplinary approach to bolster the position of solar energy in the global energy market, particularly in regions grappling with dust-related challenges.", "source_url": "https:\/\/openalex.org\/W7131736620", "pdf_url": "https:\/\/openrepository.aut.ac.nz\/bitstreams\/96943e65-ee94-479f-8e57-0648f1a0dd47\/download", "doi": "", "handle": "", "repository_name": "Tuwhera (Auckland University of Technology)", "source_record_id": "W7131736620", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 862, "title": "Definition of new parameters to develop an improved predictive model for urban heat island spatial distribution", "title_ar": "", "title_en": "Definition of new parameters to develop an improved predictive model for urban heat island spatial distribution", "author": "Victor Equere", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Nottingham", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "deep-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; deep-learning; ai-applications; predictive-analytics; ai-smart-cities", "keywords": "", "abstract": "Urban heat island (UHI) is an urban phenomenon where the inner city experiences a significant elevation in temperature, compared to the suburban areas. UHI formation and intensity have an enormous beneficial and adverse impact on the microclimate of urban spaces. Hence, there is a need to understand how several urban aspects impact this phenomenon, to enable policymakers and urban planners to propose efficient mitigation strategies to adverse the UHI effects and to benefit from its potential positive impacts. Furthermore, a robust predictive model is necessary to simulate the efficacy of a proposed mitigation or enhancement strategy before any real-world implementation. This dissertation, thus, investigates the impact of urban elevation aspects on the UHI intensity and aims to improve the prediction of the spatial distribution of the UHI by proposing two novel urban elevation parameters defined as the Elevation Factor (EF) and Terrain Factor (TF), accounting for urban superstructures and the topographical aspects of cities. The proposed parameters were integrated with other already existing urban parameters in the remote sensing subject area to train and develop predictive models with more accurate predictive performances. An artificial neural network model was developed, which resulted in further improvement of the prediction accuracy. The associated procedure for retrieving both EF and TF from point cloud LiDAR data. The urban morphology parameters used in this study were derived from two case study cities of North Park and Oak Brook towns in Illinois State, USA. The integration of TF to the training process significantly improved the model’s prediction performance of the LST for high terrain variation areas, as the average root mean square error decreased from 1.26 to 0.90⁰C and 𝑅2 increased from 0.74 to 0.81. The models were then successfully validated against the latter data of the case study cities to predict the LST for parts of the city, which were not initially included in the training process.", "source_url": "https:\/\/openalex.org\/W7217734892", "pdf_url": "https:\/\/repository.nottingham.ac.uk\/bitstreams\/abe30b91-ad4f-4f04-ab37-6858f1e8013e\/download", "doi": "", "handle": "", "repository_name": "Nottingham ePrints (University of Nottingham)", "source_record_id": "W7217734892", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 867, "title": "Temporal models for mining, ranking and recommendation in the Web", "title_ar": "", "title_en": "Temporal models for mining, ranking and recommendation in the Web", "author": "Nguyen Hoang Tu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Leibniz Universität Hannover", "university_source": "repository_host", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "nlp", "research_subdomain": "data-mining", "categories": "machine-learning; deep-learning; computer-vision; nlp; medical-ai; data-mining; predictive-analytics", "keywords": "", "abstract": "Due to their first-hand, diverse and evolution-aware reflection of nearly all areas of life, heterogeneous temporal datasets i.e., the Web, collaborative knowledge bases and social networks have been emerged as gold-mines for content analytics of many sorts. In those collections, time plays an essential role in many crucial information retrieval and data mining tasks, such as from user intent understanding, document ranking to advanced recommendations. There are two semantically closed and important constituents when modeling along the time dimension, i.e., entity and event. Time is crucially served as the context for changes driven by happenings and phenomena (events) that related to people, organizations or places (so-called entities) in our social lives. Thus, determining what users expect, or in other words, resolving the uncertainty confounded by temporal changes is a compelling task to support consistent user satisfaction. In this thesis, we address the aforementioned issues and propose temporal models that capture the temporal dynamics of such entities and events to serve for the end tasks. Specifically, we make the following contributions in this thesis: (1) Query recommendation and document ranking in the Web - we address the issues for suggesting entity-centric queries and ranking effectiveness surrounding the happening time period of an associated event. In particular, we propose a multi-criteria optimization framework that facilitates the combination of multiple temporal models to smooth out the abrupt changes when transitioning between event phases for the former and a probabilistic approach for search result diversification of temporally ambiguous queries for the latter. (2) Entity relatedness in Wikipedia - we study the long-term dynamics of Wikipedia as a global memory place for high-impact events, specifically the reviving memories of past events. Additionally, we propose a neural network-based approach to measure the temporal relatedness of entities and events. The model engages different latent representations of an entity (i.e., from time, link-based graph and content) and use the collective attention from user navigation as the supervision. (3) Graph-based ranking and temporal anchor-text mining inWeb Archives - we tackle the problem of discovering important documents along the time-span ofWeb Archives, leveraging the link graph. Specifically, we combine the problems of relevance, temporal authority, diversity and time in a unified framework. The model accounts for the incomplete link structure and natural time lagging in Web Archives in mining the temporal authority. (4) Methods for enhancing predictive models at early-stage in social media and clinical domain - we investigate several methods to control model instability and enrich contexts of predictive models at the “cold-start” period. We demonstrate their effectiveness for the rumor detection and blood glucose prediction cases respectively. Overall, the findings presented in this thesis demonstrate the importance of tracking these temporal dynamics surround salient events and entities for IR applications. We show that determining such changes in time-based patterns and trends in prevalent temporal collections can better satisfy user expectations, and boost ranking and recommendation effectiveness over time.", "source_url": "https:\/\/openalex.org\/W3014066870", "pdf_url": "https:\/\/repo.uni-hannover.de\/bitstreams\/edcbab5a-c159-4955-b94f-809b5250aea7\/download", "doi": "10.15488\/9750", "handle": "", "repository_name": "Institutional Repository of Leibniz Universität Hannover (Leibniz Universität Hannover)", "source_record_id": "W3014066870", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 871, "title": "Study of ensemble methods applied to semantic segmentation masks", "title_ar": "", "title_en": "Study of ensemble methods applied to semantic segmentation masks", "author": "João Pedro da Silva de Andrade", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Federal University of Technology – Paraná", "university_source": "repository_host", "college": "", "department": "", "country": "Brazil", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "image-segmentation", "categories": "deep-learning; computer-vision; medical-ai; cnn; image-segmentation", "keywords": "", "abstract": "Semantic segmentation is a fundamental technique for automating the detection of objects of interest in images. However, its application faces a major challenge: the need for a large amount of annotated data for training, a process that requires significant effort and time, making it unfeasible in some cases, such as applications in the healthcare field, especially those aimed at the detection of cancerous nuclei. In this context, the application of ensemble techniques can be a viable solution to improve the performance of semantic segmentation models while requiring a smaller amount of data. In view of this, this work proposes and evaluates an ensemble approach based on stacking to combine the masks generated by the U-Net and DeepLabV3+ models, combined with the VGG16, ResNet50, and MobileNetV3Large backbones. The proposal uses one or more convolutional layers to combine the masks in a way that is sensitive to the local context of the pixels, allowing the model to learn to exploit the strengths and weaknesses of each network dynamically. The objective is to generate a final mask capable of outperforming the results obtained by each individual model and also by the traditional ensemble methods evaluated. The results demonstrate that the proposed approach is statistically indistinguishable from the best static ensembles evaluated. However, it is possible that the limited number of examples available for training the stacking model limited its generalization capability.", "source_url": "https:\/\/openalex.org\/W7160572559", "pdf_url": "http:\/\/repositorio.utfpr.edu.br:8080\/jspui\/bitstream\/1\/40355\/1\/estudometodosensemblesemantica.pdf", "doi": "", "handle": "", "repository_name": "Institutional Repository of the Federal Technological University of Paraná (RIUT) (Federal University of Technology – Paraná)", "source_record_id": "W7160572559", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 875, "title": "Semantic segmentation of mammals in forest regions with U-NET", "title_ar": "", "title_en": "Semantic segmentation of mammals in forest regions with U-NET", "author": "Jhonatan Guilherme de Oliveira Cunha", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Federal University of Technology – Paraná", "university_source": "repository_host", "college": "", "department": "", "country": "Brazil", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "medical-ai", "research_subdomain": "image-segmentation", "categories": "deep-learning; computer-vision; medical-ai; cnn; image-segmentation; video-analysis", "keywords": "", "abstract": "Environmental monitoring is vital for preserving animal biodiversity, providing researchers with essential metrics on environmental health and aiding efforts to combat extinction. Mapping the fauna of forest regions is, therefore, essential for informed decision-making and effective planning. However, manual mapping is time-consuming and labor-intensive for biologists. To address this, our study utilized Deep Learning techniques to assist in detecting mammal species in the forests of northwestern Paraná, using videos provided by NAPI Biodiversidade. We implemented a semantic segmentation algorithm using the U-NET convolutional neural network to segment forest images. Our experiments demonstrated that after histogram equalization and without forest grouping, the Dice Coefficient was 79.40%, and the F1 Macro Score for animal presence classification was 86%. However, with forest grouping, there was a substantial drop in the model’s performance, highlighting difficulties in generalizing to new locations. This suggests the necessity of training location-specific models. Additionally, applying histogram equalization through contrast stretching improved the quality of the generated masks and classifications. However, the presence of false negatives indicates a need for more training data.", "source_url": "https:\/\/openalex.org\/W7155461023", "pdf_url": "http:\/\/repositorio.utfpr.edu.br:8080\/jspui\/bitstream\/1\/35300\/1\/segmentacaosematicamamiferosunet.pdf", "doi": "", "handle": "", "repository_name": "Institutional Repository of the Federal Technological University of Paraná (RIUT) (Federal University of Technology – Paraná)", "source_record_id": "W7155461023", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 877, "title": "Semantic segmentation of image sequences using a spatio-temporal U-Net", "title_ar": "", "title_en": "Semantic segmentation of image sequences using a spatio-temporal U-Net", "author": "Manuel Danner", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "TU Wien", "university_source": "repository_host", "college": "", "department": "", "country": "Austria", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "medical-ai", "research_subdomain": "image-segmentation", "categories": "deep-learning; computer-vision; medical-ai; image-segmentation; video-analysis", "keywords": "", "abstract": "In biomedical research, detailed structure of tissues, cells, organelles and macromolecular complexes is investigated with electron microscopy (EM) images. On this account, large amounts of high resolution images from biological and clinical specimens exist. As a result, there is a need for computer assisted tools that can provide a cost effective solution for disease diagnostics. This thesis illustrates a novel elastic image transformation method called Elastic Gradient Transformation (EGT), which uses the image gradient to generate realistic looking deformations of cell structures. The novel EGT method helps our neural network to generalize on little cell datasets (like the ISBI 2012 dataset), without overfitting. The U-Net architecture by O. Ronneberger, P. Fischer and T. Brox, is adapted to contain an additional encoder path. The proposed network is called SiamU-Net, and takes two sequential images t and t+1 as input. The output is a class probability map of image t+1. It is important that both encoder paths do not share weights, and are fused together in the latent space of the network. The single decoding path uses skip connections from the encoding path of image t+1 to generate an improved up-sampling. To evaluate the adaptation, both U-Net and SiamU-Net use the novel elastic gradient transformation method and participate in the ISBI 2012 challenge. To highlight the impact of temporal image information on the two networks, a comparison of both networks is made on a video dataset called DAVIS 2016. At the ISBI 2012 challenge, the proposed SiamU-Net with the EGT method is placed at rank 31, while the original U-Net is placed at rank 61, out of 223 participants. On the Davis 2016 challenge, the SiamU-Net achieves a 0.0776 point higher Jaccard Index than the U-Net architecture, which proves the advantage of adapting the U-Net with an additional encoder path.", "source_url": "https:\/\/openalex.org\/W3087797547", "pdf_url": "", "doi": "10.34726\/hss.2020.66920", "handle": "", "repository_name": "reposiTUm (TU Wien)", "source_record_id": "W3087797547", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 879, "title": "Semantic segmentation of agricultural crop areas via two-stage U-Net", "title_ar": "", "title_en": "Semantic segmentation of agricultural crop areas via two-stage U-Net", "author": "Walysson Carlos dos Santos Oliveira", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "computer-vision", "research_subdomain": "image-segmentation", "categories": "machine-learning; computer-vision; ai-applications; image-segmentation; ai-agriculture", "keywords": "", "abstract": "The agribusiness tax is mainly levied on the production of agricultural crops. To reduce tax evasion in agribusiness, it is possible to monitor the development of plantations through the analysis of satellite images. For this, we can apply Machine Learning techniques to satellite images to segment the planted area, and the area, in turn, can be used to estimate the production of monitored plantations. This work aims to solve the first stage of the problem, the Segmentation of the Planted Area. For this, we developed a machine learning architecture for segmentation of plantation areas, the Two-stage U-net. In addition, the work also included the creation of a satellite image dataset with annotations for the segmentation of plantation areas. We trained the proposed model and we adjusted its hyperparameters considering the U-net Encoder, the Optimizer, the Loss Function, and the Batch Size of images. We selected the fitted model that performed best in tests with Hyperopt and GridSearch. The results in mIoU of the Two-stage U-net were superior to the results of other architectures used in similar works.", "source_url": "https:\/\/openalex.org\/W7120464696", "pdf_url": "", "doi": "", "handle": "", "repository_name": "LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas)", "source_record_id": "W7120464696", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 887, "title": "Interpretability of a Deep Learning Model for Semantic Segmentation : Example of Remote Sensing Application", "title_ar": "", "title_en": "Interpretability of a Deep Learning Model for Semantic Segmentation : Example of Remote Sensing Application", "author": "Adrianna Janik", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "KTH Royal Institute of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "computer-vision", "research_subdomain": "image-segmentation", "categories": "machine-learning; deep-learning; computer-vision; responsible-ai; ai-applications; predictive-analytics; image-segmentation; xai; fairness; ai-smart-cities", "keywords": "", "abstract": "Understanding a black-box model is a major problem in domains that relies on model predictions in critical tasks. If solved, can help to evaluate the trustworthiness of a model. This thesis proposes a user-centric approach to black-box interpretability. It addresses the problem in semantic segmentation setting with an example of humanitarian remote sensing application for building detection. The question that drives this work was, Can existing methods for explaining black-box classifiers be used for a deep learning semantic segmentation model? We approached this problem with exploratory qualitative research involving a case study and human evaluation. The study showed that it is possible to explain a segmentation model with adapted methods for classifiers but not without a cost. The specificity of the model is likely to be lost in the process. The sole process could include introducing artificial classes or fragmenting image into super-pixels. Other approaches are necessary to mitigate identified drawback. The main contribution of this work is an interactive visualisation approach for exploring learned latent space via a deep segmenter, named U-Net, evaluated with a user study involving 45 respondents. We developed an artefact (accessible online) to evaluate the approach with the survey. It presents an example of this approach with a real-world satellite image dataset. In the evaluation study, the majority of users had a computer science background (80%), including a large percentage of users with machine learning specialisation (44.4% of all respondents). The model distinguishes rurality vs urbanization (58% of users). External quantitative comparison of building densities of each city concerning the location in the latent space confirmed the later. The representation of the model was found faithful to the underlying model (62% of users). Preliminary results show the utility of the pursued approach in the application domain. Limited possibility to present complex model visually requires further investigation.", "source_url": "https:\/\/openalex.org\/W3011114404", "pdf_url": "", "doi": "", "handle": "", "repository_name": "KTH Publication Database DiVA (KTH Royal Institute of Technology)", "source_record_id": "W3011114404", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 890, "title": "Iris segmentation", "title_ar": "", "title_en": "Iris segmentation", "author": "Vishal Ramesh", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Czech Republic", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "medical-ai", "research_subdomain": "image-segmentation", "categories": "deep-learning; computer-vision; medical-ai; image-segmentation", "keywords": "", "abstract": "Accurate iris image segmentation is crucial to a range of proposed medical diagnosis and treatment systems. Previous models have worked well with healthy eye images but do not generalize to diseased eye images. We work with a dataset where many subjects have eye diseases or deformities. We analyse the performance of the U-Net, a deep learning architecture for semantic segmentation. Our model was trained on a hand-annotated dataset and tuned to generalise on unseen images. Our model achieves a pixel accuracy of 0.8913 on the test set with a relatively short training time. 1", "source_url": "https:\/\/openalex.org\/W7135536303", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Digital Repository (National Repository of Grey Literature)", "source_record_id": "W7135536303", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 894, "title": "Multidimensional Transfer Learning in Natural Language Processing: Domain, Task, Language, and Expertise Adaptation", "title_ar": "", "title_en": "Multidimensional Transfer Learning in Natural Language Processing: Domain, Task, Language, and Expertise Adaptation", "author": "Nghia Ngo", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Oregon", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; nlp; medical-ai; transfer-learning; predictive-analytics; llm", "keywords": "", "abstract": "Transfer learning in Natural Language Processing (NLP) mirrors a fundamental human cognitive ability: applying prior knowledge to learn new skills efficiently. Rather than training models from scratch for every new task, domain, or language, transfer learning enables systems to use knowledge from previous learning experiences. This capability is essential for addressing two critical challenges in modern NLP: the scarcity of labeled data affecting the vast majority of the world’s languages and specialized domains, and the prohibitive computational cost of training large-scale models from scratch. This dissertation presents a comprehensive investigation of transfer learning in NLP, organized around two distinct phases. The first phase, Adaptive Transfer Learning, addresses horizontal knowledge transfer across contexts through three research directions: unsupervised domain adaptation, where models must transfer across data distributions without target labels; task adaptation for structured prediction, where multiple interdependent tasks such as joint information extraction must be learned and transferred simultaneously across domains; and cross-lingual transfer, where knowledge from high-resource languages is shared to low-resource languages across typologically diverse language families. The second phase, Specialization Transfer Learning, addresses the emerging challenge of vertically refining general-purpose large language models for reliable deployment in high-stakes expert domains. This phase focuses on developing comprehensive evaluation frameworks for Retrieval-Augmented Generation systems in the medical domain, assessing critical dimensions beyond standard accuracy such as knowledge sufficiency, multi-document integration, and robustness to noisy or misleading retrieved content. Together, these contributions advance the understanding of how NLP models can be systematically transferred and specialized across domains, tasks, languages, and levels of expertise, spanning the full spectrum from data-limited to data-engineered paradigms. This dissertation includes both previously published and co-authored material.", "source_url": "https:\/\/openalex.org\/W7165425966", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Scholars' Bank (University of Oregon)", "source_record_id": "W7165425966", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 897, "title": "Transfer learning for language model adaptation", "title_ar": "", "title_en": "Transfer learning for language model adaptation", "author": "Bari M. Saiful", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Nanyang Technological University", "university_source": "repository_host", "college": "", "department": "", "country": "Singapore", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; transfer-learning; self-semi-supervised; predictive-analytics; llm; sentiment-analysis; machine-translation", "keywords": "", "abstract": "Language is the pathway to democratize the boundary of land and culture. Bridging the gap between languages is one of the biggest challenges of Artificial Intelligent (AI) systems. The current success of AI systems is dominated by the supervised learning paradigm where gradient-based learning algorithms (i.e., SGD, Adam) are designed to optimize complex high-dimensional planes. These algorithms learn from statistical observations that are typically collected with the intention of a specific task (i.e., product review, sentiment analysis). The use of task-dependent samples makes the learning procedure tedious as it requires manually annotated data. On the contrary, without having a sufficient amount of samples to represent the distribution, deep learning models tend to suffer from the lack of robustness. Due to the natural puzzle of randomness, not all sets of observations are observed in the data collection procedure, thus creating Out-of-Distribution (OOD) problem in learning algorithms. In the search of a generic task agnostic distribution, a large collection of text over various domains can be considered as - Standard Natural Text Distribution (SNTD). The generic idea of Transfer Learning for Natural Language Processing (NLP) is to utilize the SNTD knowledge for any other task-dependent training. Learning SNTD, followed by a task-adaptation method with a smaller volume of annotated data has yielded state-of-the-art (SOTA) results in various supervised NLP tasks. However, annotated data for every task in every language is rare. In language models, there are many kinds of distributional variance. One of the most common ways that distributional variance is encoded into a language model is when it is trained with mono-lingual text and learned dis-jointly. Word embeddings produced from these language models are then used as pre-trained embedding vectors to adapt to a downstream task. We propose adversarial training to project two monolingual distributions into the same space and then improve the robustness of the model by augmented fine-tuning with parameter sharing. By projecting mono-lingual language distribution into the same cross-lingual space, the distribution of the language become aware of each other. These projected distribution are semantically aware of each other in the latent space. Thus when we train one distribution the other distribution automatically adapt to the training data, making it easier to transfer (exchange) knowledge. On top of that, the proposed novel self-training architecture improves cross-lingual transfer by a large margin. Then, we focus on the jointly trained multi-lingual language model, where there is no pre-dominant distributional variance. In the multi-lingual language model, we put more focus on downstream task adaptation. We found that semi-supervised learning with pseudo-augmented data from a pre-trained language model can greatly improve the performance of tasks further down the line. In the end, we introduce a new novel data augmentation framework that uses the neighboring (vicinal) samples of the original training data without explicitly using any parallel text corpora or machine translation system. Our proposed method performs simultaneous self-training with data augmentation and unsupervised sample selection. It also proposes curriculum strategies for different domain samples. With extensive experiments on three different cross-lingual tasks spanning many language pairs, we have demonstrated the effectiveness of our proposed method. While all of the work done above focused on improving task adaptation in multiple languages without supervision, we further investigate how adding a few samples affects multi-lingual task adaptation. To this end, we leverage a small number of support samples from each language and propose an inference-time transductive nearest neighbor-based approach which utilizes the entropy of the query samples for predictions. We show that our proposed method outperforms full model\/full head fine-tuning as well as cross-task fine-tuning. We also show impressive performance gain (~37x) in computational costs for the full inference prediction. However, as the language models grow bigger, it becomes increasingly difficult to perform efficient inferences, especially for multiple tasks. The jointly optimized multi-lingual distribution helps transfer knowledge from a high-resource language to a low-resource language. While working on transductive nearest neighbor inference, we observe that language models are significantly prone to task distributions. Unless we use an extremely large language model (>100B), a model used for a specific task adaptation cannot be used for another task. In this dissertation, our final proposed method addresses this issue by multitask prompted learning.", "source_url": "https:\/\/openalex.org\/W4386409617", "pdf_url": "https:\/\/dr.ntu.edu.sg\/bitstream\/10356\/169892\/2\/sbmaruf_thesis.pdf", "doi": "10.32657\/10356\/169892", "handle": "", "repository_name": "DR-NTU (Nanyang Technological University)", "source_record_id": "W4386409617", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 899, "title": "Deep Learning Domain Adaptation in Brain MRI: Investigating Motion Mitigation in Adult and Neonatal Scans", "title_ar": "", "title_en": "Deep Learning Domain Adaptation in Brain MRI: Investigating Motion Mitigation in Adult and Neonatal Scans", "author": "Saad Bin Ashraf", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; transfer-learning; image-segmentation; medical-imaging", "keywords": "", "abstract": "Magnetic Resonance Imaging (MRI) is a vital non-invasive tool for high-resolution brain imaging, yet its long acquisition time frequently results in patient movement leading to motion artifacts, particularly among non-compliant patients like newborns. While Deep Learning (DL) models have emerged as powerful retrospective solutions for motion mitigation, their performance is expected to degrade when applied to data from different scanners, protocols, or patient populations. This thesis investigates the generalizability of DL-based motion correction across heterogeneous adult and neonatal brain MRI domains. The research is conducted through three primary aims. First, the study evaluates the domain generalization of a 3D U-Net model across multi-center adult datasets - IXI, Calgary-Campinas, and OASIS-3, comprising 11 imaging centers with diverse scanner vendors and field strengths. Results indicate that models generalize effectively to unseen adult datasets from various centers, while superior performance is achieved when training incorporates broader intensity distributions. Second, the thesis explores cross-domain transfer learning by applying adult-trained models to the neonatal domain, utilizing the dHCP dataset, which contains real motion from neonatal subjects. Results indicate that adult-pretrained models learn domain-agnostic structural priors, allowing them to reconstruct neonatal brain anatomy despite significant shifts in contrast, size, and motion characteristics. Crucially, fine-tuning with as few as three neonatal scans was shown to substantially improve the mitigation of neonatal-specific artifacts, demonstrating an efficient adaptation strategy for data-scarce domains. Third, the study critically assesses the reliability of common image quality metrics (IQMs). The analysis reveals that traditional metrics are often insensitive to subtle motion and can even yield paradoxical results, where motion-corrupted images appear higher in quality than corrected ones. Overall, this work highlights the potential for deploying robust motion-mitigation frameworks in diverse clinical settings. By leveraging transfer learning, this thesis provides a foundation for more reliable and scalable deep learning–based motion correction in pediatric and adult neuroimaging.", "source_url": "https:\/\/openalex.org\/W7140225405", "pdf_url": "", "doi": "10.11575\/prism\/51192", "handle": "", "repository_name": "", "source_record_id": "W7140225405", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 908, "title": "Domain adaptive learning with disentangled features", "title_ar": "", "title_en": "Domain adaptive learning with disentangled features", "author": "Xingchao Peng", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Boston University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "transfer-learning", "categories": "machine-learning; deep-learning; computer-vision; robotics; transfer-learning", "keywords": "", "abstract": "Recognizing visual information is crucial for many real artificial-intelligence-based applications, ranging from domestic robots to autonomous vehicles. However, the success of deep learning methods on visual recognition tasks is highly dependent on access to large-scale labeled datasets, which are expensive and cumbersome to collect. Transfer learning provides a way to alleviate the burden of annotating data, which transfers the knowledge learned from a rich-labeled source domain to a scarce-labeled target domain. However, the performance of deep learning models degrades significantly when testing on novel domains due to the presence of domain shift. To tackle the domain shift, conventional domain adaptation methods diminish the domain shift between two domains with a distribution matching loss or adversarial loss. These models align the domain-specific feature distribution and the domain-invariant feature distribution simultaneously, which is sub-optimal towards solving deep domain adaptation tasks, given that deep neural networks are known to extract features in which multiple hidden factors are highly entangled. This thesis explores how to learn effective transferable features by disentangling the deep features. The following questions are studied: (1) how to disentangle the deep features into domain-invariant and domain-specific features? (2) how would feature disentanglement help to learn transferable features under a synthetic-to-real domain adaptation scenario? (3) how would feature disentanglement facilitate transfer learning with multiple source or target domains? (4) how to leverage feature disentanglement to boost the performance in a federated system? To address these needs, this thesis proposes deep adversarial feature disentanglement: a class\/domain identifier is trained on the labeled source domain and the disentangler generates features to fool the class\/domain identifier. Extensive experiments and empirical analysis demonstrate the effectiveness of the feature disentanglement method on many real-world domain adaptation tasks. Specifically, the following three unsupervised domain adaptation scenarios are explored: (1) domain agnostic learning with disentangled representations, (2) unsupervised federated domain adaptation, (3) multi-source domain adaptation.", "source_url": "https:\/\/openalex.org\/W3137379704", "pdf_url": "", "doi": "", "handle": "", "repository_name": "OpenBU (Boston University)", "source_record_id": "W3137379704", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 912, "title": "Adversarial Domain Adaptation for Sensor Networks", "title_ar": "", "title_en": "Adversarial Domain Adaptation for Sensor Networks", "author": "Francisco Tuna de Andrade", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Portugal", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "deep-learning", "research_subdomain": "rnn", "categories": "machine-learning; deep-learning; generative-ai; transfer-learning; rnn", "keywords": "", "abstract": "With the recent surge of big data, manually annotating datasets has become an impossible task. Asa result, domain adaptation has gained more importance than ever, since it allows the transfer oflearning from a labeled dataset to an unlabeled one. That is even truer for sensor networks, wherethe domain shifts between different sensors are usually substantial and models developed with aset of source sensors fail to generalize well to new target sensors.Adversarial neural networks, which were introduced in 2014 in the scope of generative adver-sarial neural networks, are a promising tool to address the domain adaptation problem. Despitebeing a recent idea, adversarial networks have already led to tremendous progress in multiple areasof the field. In this dissertation, we propose a new model for domain adaptation, specifically tunedfor sensor networks, that consists of two adversarial Long short-term memory (LSTM) networks.One that seeks to discriminate between source and target domains and another one that performsthe learning task. The fact that we take the temporal nature of the data into account by using anLSTM network for the discriminating task is a novel idea. We evaluate the performance of ourmodel by conducting extensive experiments and comparing it with other state of the art methods.Besides the proposed model, we include an exhaustive survey of domain adaptation methodsand an experimental analysis of their efficiency in different datasets. Additionally, we also discussseveral open lines of research in the field of domain adaptation that can serve as a guide for futurework in this area.", "source_url": "https:\/\/openalex.org\/W3048615177", "pdf_url": "", "doi": "10.34626\/fdkt-sv10", "handle": "", "repository_name": "Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT)", "source_record_id": "W3048615177", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 913, "title": "Domain Adaptation and Domain Generalization with Representation Learning", "title_ar": "", "title_en": "Domain Adaptation and Domain Generalization with Representation Learning", "author": "Muhammad Ghifary", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "New Zealand", "governorate": "", "publication_year": 2016, "language": "en", "research_domain": "computer-vision", "research_subdomain": "video-analysis", "categories": "machine-learning; deep-learning; computer-vision; responsible-ai; transfer-learning; cnn; image-classification; video-analysis; fairness", "keywords": "", "abstract": "Machine learning has achieved great successes in the area of computer vision, especially in object recognition or classification. One of the core factors of the successes is the availability of massive labeled image or video data for training, collected manually by human. Labeling source training data, however, can be expensive and time consuming. Furthermore, a large amount of labeled source data may not always guarantee traditional machine learning techniques to generalize well; there is a potential bias or mismatch in the data, i.e., the training data do not represent the target environment. To mitigate the above dataset bias\/mismatch, one can consider domain adaptation: utilizing labeled training data and unlabeled target data to develop a well-performing classifier on the target environment. In some cases, however, the unlabeled target data are nonexistent, but multiple labeled sources of data exist. Such situations can be addressed by domain generalization: using multiple source training sets to produce a classifier that generalizes on the unseen target domain. Although several domain adaptation and generalization approaches have been proposed, the domain mismatch in object recognition remains a challenging, open problem – the model performance has yet reached to a satisfactory level in real world applications. The overall goal of this thesis is to progress towards solving dataset bias in visual object recognition through representation learning in the context of domain adaptation and domain generalization. Representation learning is concerned with finding proper data representations or features via learning rather than via engineering by human experts. This thesis proposes several representation learning solutions based on deep learning and kernel methods. This thesis introduces a robust-to-noise deep neural network for handwritten digit classification trained on “clean” images only, which we name Deep Hybrid Network (DHN). DHNs are based on a particular combination of sparse autoencoders and restricted Boltzmann machines. The results show that DHN performs better than the standard deep neural network in recognizing digits with Gaussian and impulse noise, block and border occlusions. This thesis proposes the Domain Adaptive Neural Network (DaNN), a neural network based domain adaptation algorithm that minimizes the classification error and the domain discrepancy between the source and target data representations. The experiments show the competitiveness of DaNN against several state-of-the-art methods on a benchmark object dataset. This thesis develops the Multi-task Autoencoder (MTAE), a domain generalization algorithm based on autoencoders trained via multi-task learning. MTAE learns to transform the original image into its analogs in multiple related domains simultaneously. The results show that the MTAE’s representations provide better classification performance than some alternative autoencoder-based models as well as the current state-of-the-art domain generalization algorithms. This thesis proposes a fast kernel-based representation learning algorithm for both domain adaptation and domain generalization, Scatter Component Analysis (SCA). SCA finds a data representation that trades between maximizing the separability of classes, minimizing the mismatch between domains, and maximizing the separability of the whole data points. The results show that SCA performs much faster than some competitive algorithms, while providing state-of-the-art accuracy in both domain adaptation and domain generalization. Finally, this thesis presents the Deep Reconstruction-Classification Network (DRCN), a deep convolutional network for domain adaptation. DRCN learns to classify labeled source data and also to reconstruct unlabeled target data via a shared encoding representation. The results show that DRCN provides competitive or better performance than the prior state-of-the-art model on several cross-domain object datasets.", "source_url": "https:\/\/openalex.org\/W2516289717", "pdf_url": "https:\/\/openaccess.wgtn.ac.nz\/articles\/thesis\/Domain_Adaptation_and_Domain_Generalization_with_Representation_Learning\/17014700\/1\/files\/31470596.pdf", "doi": "10.26686\/wgtn.17014700", "handle": "", "repository_name": "Open Access Victoria University of Wellington | Te Herenga Waka (Figshare)", "source_record_id": "W2516289717", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 914, "title": "Learning to recognize objects by adaptive knowledge transfer", "title_ar": "", "title_en": "Learning to recognize objects by adaptive knowledge transfer", "author": "Qingyi Tao", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Nanyang Technological University", "university_source": "repository_host", "college": "", "department": "", "country": "Singapore", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; transfer-learning; object-detection; medical-imaging", "keywords": "", "abstract": "When humans learn new knowledge and skills, we can naturally transfer them to other domains. Along with the learning procedures, we learn knowledge and skills for certain tasks and transfer them to similar tasks; we also can use the old knowledge to facilitate the learning of new knowledge. While effective knowledge transfer is a congenital and important learning ability of humans, it is not easy for machine learning mechanisms to adopt the ability of knowledge transfer. In recent years, there are plenty of works studying transfer learning in deep learning. There are still some practical challenges that remain undiscovered, especially under different problem settings encountered in real situations. In this thesis, we explore how to adopt knowledge transfer mechanisms in deep learning approaches in several practical scenarios. Four different works are proposed to study knowledge transfer across different domains and tasks via domain adaptation and model transfer. In particular, we study the web knowledge transfer for object detection task by adapting the web data to the real target dataset, which aims at reducing the human annotation effort for training object detector. In the incremental learning scenario, we study the cross-utilization of the old and new knowledge to overcome the catastrophic forgetting during the incremental and progressive learning process. Lastly, we explore transfer learning in the medical imaging domain by transferring the model pre-trained on normal images. Overall, the major contributions are summarized as follows: - A web knowledge transfer method to enhance the learning of weakly supervised object detection. The proposed method includes an effective web data collection pipeline and a curriculum learning scheme to achieve more effective model optimization during multi-instance learning. - An annotation-effective object detection method by adapting web data to the target data for object detection. This work attempts to learn an object detector from web supervision by adversarial domain adaptation. - An incremental learning scheme that adapts an old model to a new model without forgetting the old knowledge. A systematic study is performed to explore different class incremental methods. Furthermore, we propose a graph-based method to mine the old sample forgettability along with the training of the new tasks, and dynamically select samples that are more forgettable to overcome the catastrophic forgetting. - A lesion detection method for 3D CT images by utilizing model weights that are pre-trained on normal 2D RGB images. Furthermore, an attention-based feature aggregation method is proposed to adaptively transfer the information from neighboring slices to the key slices for more discriminative representation. Through this thesis, we demonstrate three different paradigms of knowledge transfer, including (1) the cross-domain knowledge transfer for adapting web data to an application with real unconstrained data, (2) the continual knowledge transfer from old tasks to new tasks without forgetting the old knowledge, and (3) the model transfer from the normal image domain to the medical domain. Under several practical tasks, the experiments are conducted to demonstrate the effectiveness of the proposed knowledge transfer approaches.", "source_url": "https:\/\/openalex.org\/W3177101414", "pdf_url": "https:\/\/dr.ntu.edu.sg\/bitstream\/10356\/151507\/4\/thesis_oneside_2021_taoqingyi.pdf", "doi": "10.32657\/10356\/151507", "handle": "", "repository_name": "DR-NTU (Nanyang Technological University)", "source_record_id": "W3177101414", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 924, "title": "Artificial Intelligence for Islamic Ontology Linking Qur’an and Hadith Resources", "title_ar": "", "title_en": "Artificial Intelligence for Islamic Ontology Linking Qur’an and Hadith Resources", "author": "Ibtisam Khalaf Alshammari", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "deep-learning; nlp; ai-applications; transformers; llm; arabic-nlp; ai-education", "keywords": "", "abstract": "The primary objective of this thesis is to integrate various Islamic resources, specifically the Qur’an and Hadith, building upon existing academic research. Although previous works have produced valuable contributions, they remain uncoordinated and heterogeneous, raising significant challenges for interoperability and integration. This thesis addresses these limitations by unifying these inconsistent resources into a coherent and comprehensive Islamic ontology, ensuring a more accurate and structured representation of Islamic knowledge and facilitating wider computational applications. The project employs a multi-stage methodology. A workflow combining RML, Cellfie plugin, and SDM-RDFizer interpreter is utilised to integrate Qur’anic resources, enabling data transformation and alignment. This framework successfully maps the data correctly, demonstrating the effectiveness of the Cellfie plugin tool. For the Hadith corpus, knowledge extraction approaches are applied to the LK-Hadith corpus. First experiment with Arabic named entity recognition (ANER) using a limited number of BERT-based Arabic pretrained language models (PLMs) employed on the Sahih Albukhari book produces insufficient results for semantic integration, leading to the adoption of topic modelling methods. The topic modelling and extraction involves Latent Dirichlet Allocation (LDA), Non-negative Matrix Factorization (NMF), BERTopic with Arabic PLMs, and GPT-4. This experiment presented a bilingual Hadith_Teaching_Topics (HTT) dataset. Then, the semantic relationships between Qur’anic and Hadith topics are evaluated through the Quran_Hadith_Dataset (QH_Dataset) using multiple large language models (LLMs), enriching the construction of the Related Quran-Hadith Topics (RQHT) ontology. The ontology is subsequently extended to incorporate the QuranOntology, the entire six Hadith books derived from the LK-Hadith corpus, and their derived topics based on the HTT dataset. This thesis advances the field by establishing a methodological framework for integrating heterogeneous Islamic textual resources by involving multiple data transformations, knowledge extraction, and ontology engineering techniques. Within this framework, the Qur’anic resources are integrated to enhance the unification process, the bilingual HTT dataset is developed to augment the low-resource Arabic annotation and support the final integrated ontology, and the RQHT ontology is introduced to formally represent the semantic relationships between Qur’an and Hadith topics. Collectively, these contributions boost the unification process of Qur’an and Hadith knowledge within a comprehensive and logically consistent semantic Islamic ontology. Evaluation through quantitative and qualitative measures determines the Islamic ontology’s logical consistency, representational completeness, and suitability for advanced computational applications. This research provides a scalable semantic framework to support more knowledge representation and reasoning across core Islamic texts.", "source_url": "https:\/\/openalex.org\/W7134994163", "pdf_url": "https:\/\/etheses.whiterose.ac.uk\/id\/eprint\/38206\/1\/Artificial_Intelligence_for_Islamic_Ontology_Linking_Qur_an_and_Hadith_Resources.pdf", "doi": "", "handle": "", "repository_name": "White Rose eTheses Online (University of Leeds, The University of Sheffield, University of York)", "source_record_id": "W7134994163", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 929, "title": "Geospatial AI-based approach to assess the spatiotemporal suitability of onshore wind-solar farms in Iraq", "title_ar": "", "title_en": "Geospatial AI-based approach to assess the spatiotemporal suitability of onshore wind-solar farms in Iraq", "author": "Sachit, Mourtadha Sarhan Almushattat", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Universiti Putra Malaysia", "university_source": "repository_host", "college": "", "department": "", "country": "Malaysia", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "responsible-ai", "research_subdomain": "xai", "categories": "machine-learning; responsible-ai; ai-applications; xai; ai-agriculture; ai-smart-cities", "keywords": "", "abstract": "Nowadays, Renewable Energy (RE), such as wind and solar, play a vital role in meeting the increasing demand for electricity and ensuring a low-carbon future. Unfortunately, the efficiency of wind and solar plants depends on the spatially and temporally fluctuating nature of their renewable resources. These fluctuations have prompted planners and decision-makers to shift toward hybridizing such energies in an effort to make electricity generation more stable. Wind-solar hybrid plants have posed new challenges in evaluating suitable sites that meet the requirements of both types of energy. As a result, site selection models for hybrid power systems (wind-solar) have received wide attention in recent years as being a critical planning problem that needs accurate decision-making. In the existing literature, the site suitability assessment of dual-energy systems (windsolar) is frequently addressed as a Spatial Decision-Making (SDM) problem involving numerous climatic, economic, and environmental criteria. In most cases, the input factors are considered constant. However, most of these factors, such as wind speed and solar radiation, change over time. Besides, criteria are often subjectively or objectively weighted, generating biased and non-generalizable solutions. To overcome these challenges, the current research aims to develop a SpatioTemporal Decision-Making (STDM) model based on Geospatial Artificial Intelligence (GeoAI) to locate onshore wind-solar hybrid plants. The presented model seeks to fill in the existing gaps by considering the dynamic nature of decision criteria and formulating novel, more reliable global weights. To achieve the research goal, a four-stage methodology was drawn. A system of spatial evaluation criteria was first designed based on literature statistics and expert judgments supported by content validity analysis. Second, eXplainable Artificial Intelligence (XAI) was introduced to formulate novel global weights for those criteria. In this context, global geospatial data for 13 conditioning factors were collected, and 55,619 inventory samples of wind and solar stations worldwide were prepared to train three machine learning (ML) algorithms, namely Random Forest (RF), Support Vector Machine (SVM), and Multi-Layer Perceptron (MLP). SHapley Additive exPlanations (SHAP) algorithm was applied to interpret the results of the superior model and elicit the criteria weights. In a separate third step, a new temporal criterion was developed based on temporal complementarity assessment between wind and solar resources. Pearson’s correlation coefficients were utilized to explore the synergistic patterns between the time-series dataset of wind speed and solar radiation across Iraqi territory. Finally, spatial, temporal, and exclusion criteria were incorporated along with AI weights into a unified STDM model using multiple overlay analyses in a GIS environment. The outcomes of the criteria design demonstrated that 13 factors, with an excellent Kappa value of 0.76, would form valuable content for evaluating the spatial suitability of wind and solar farms. The results indicated the superiority of the RF algorithm in both wind and solar modeling, with an overall accuracy of 90% and 89%, a kappa coefficient of 0.79 and 0.78, and an area under the curve of 0.96 and 0.95, respectively. The XAIbased importance analysis revealed higher weights for technical and economic criteria than for other environmental and social criteria. Top weights were given to the criteria of wind speed and proximity to cities (0.373 and 0.149, respectively) in locating wind farms, and the criteria of proximity to cities and air temperature (0.180 and 0.149, respectively) in locating solar farms. The southwestern regions and some eastern parts of Iraq exhibited significant temporal synergistic patterns spanning more than 6 months of the year that influence spatial decision-making. Our spatiotemporal model identified three hotspots over Iraq—South Dhi-Qar, West Diyala, and East Wasit—with a total area of 3,632 km2. The hotspots revealed exceptional suitability scores exceeding 0.8, meeting both the spatial and temporal constraints. The reported spots have the technical potential to generate electricity from wind turbines and solar PV cells at rates of 11.88– 12.58 MW and 81.31–89.51 MW, respectively. Overall, this research has led to the development of a new GeoAI-based STDM model that can find the best places to put wind-solar systems with great accuracy and consistency. Analyses of transferability, sensitivity, and uncertainty show that the GeoAI-based STDM model is more reliable than one-dimensional solutions. The advantage of the proposed model is not only to identify the technically, economically, and environmentally appropriate sites but also to ensure that they boast temporal synergistic patterns between renewables for stable power supplies around the clock.", "source_url": "https:\/\/openalex.org\/W7112578286", "pdf_url": "http:\/\/psasir.upm.edu.my\/id\/eprint\/115915\/1\/115915.pdf", "doi": "", "handle": "", "repository_name": "Universiti Putra Malaysia Institutional Repository (Universiti Putra Malaysia)", "source_record_id": "W7112578286", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 931, "title": "Title: Employing Earth Observations and Artificial Intelligence to Address Key Global Environmental Challenges in Service of the SDGs", "title_ar": "", "title_en": "Title: Employing Earth Observations and Artificial Intelligence to Address Key Global Environmental Challenges in Service of the SDGs", "author": "Wenzhao Li", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Chapman University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "deep-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; deep-learning; ai-applications; predictive-analytics; ai-smart-cities", "keywords": "", "abstract": "Earth Observation (EO) data provides the capability to integrate data from multiple sources and helps to produce more relevant, frequent, and accurate information about complex processes. EO, empowered by methodologies from Artificial Intelligence (AI), supports various aspects of the UN’s Sustainable Development Goals (SDGs). This dissertation presents author’s major studies using EO to fill in knowledge gaps and develop methodologies and cloud-based applications in selected SDGs, including SDG 6 (Clean Water and Sanitation), SDG 11 (Sustainable Cities and Communities), SDG 14 (Life below Water) and SDG 15 (Life on Land). For SDG 6, the study focuses on spatiotemporal water recharge patterns and interconnections between variables in the Nile watershed countries, highlighting the EO’s potential to implement transboundary water cooperation (SDG 6.5.2). For SDG 11, the studies focus on the indicators of urbanization (SDG 11.a.1 and 11.1.1) and air quality (SDG 11.6.2). Utilizing EO datasets, the annual geographical and population-weighted PM 2.5 levels in Egypt are analyzed in 18 years. The Random Decision Forest is developed to predict the PM10 levels in Cairo. Additionally, a cloud-based impervious surface classifier based on fully convolution neural networks (FCNN) is trained to monitor the urbanization in five Nile basin cities between 2013 and 2019. For SDG 14, the studies focus on marine pollution (SDG 14.1.1) through monitoring and exploring the atmospheric and meteorological factors regulating algal blooms, as well as its primary productivity in the Red Sea. For SDG 15, the mangrove changes over the Western Gulf region is quantified using multi-indices and machine learning models to protect the local threatened species (SDG 15.5.1). The MENA is investigated concerning changes in land coverage using the harmonic analysis of vegetation index (SDG 15.1.1). Finally, the dissertation also summarizes the SDG researches the author involved, including 1) air pollutant in the MENA region; 2) dust impact on rainfall 3) aerosols’ impact on snowfall; 4) oil spill modeling to help proper mitigation; 5) vegetation in the sea turtle nesting Islands; and 6) EO of Mediterranean seagrass Posidonia Oceanica.", "source_url": "https:\/\/openalex.org\/W3002647053", "pdf_url": "", "doi": "10.36837\/chapman.000115", "handle": "", "repository_name": "Chapman University Digital Commons (Chapman University)", "source_record_id": "W3002647053", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 933, "title": "On the Safety Alignment of Large Language Models", "title_ar": "", "title_en": "On the Safety Alignment of Large Language Models", "author": "Mengxuan Hu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; nlp; medical-ai; responsible-ai; ai-applications; transfer-learning; llm; fairness; ai-education; ai-cybersecurity", "keywords": "", "abstract": "Large language models (LLMs) are increasingly deployed in high-stakes settings such as finance, healthcare, and education, where failures in safety and fairness can cause real harm. In this dissertation, we advance LLM trustworthiness by (i) benchmarking how LLMs introduce explicit risks within the RAG pipeline, (ii) developing red-teaming attacks that expose latent vulnerabilities even under seemingly “safe” conditions, and (iii) proposing robust defenses that strengthen alignment in practice. First, we study Retrieval-Augmented Generation (RAG), a widely used strategy to reduce hallucinations and improve domain grounding, and show that its benefits are not a free lunch. We propose a practical three-level threat model based on user awareness of fairness, which induces different degrees of “fairness censorship” in external dataset. Across uncensored, partially censored, and fully censored datasets, we find that fairness alignment can be undermined through RAG alone, without any fine-tuning, and that biased outputs can persist even when the external data is ostensibly unbiased. These results reveal fundamental limitations of current alignment when models are coupled with retrieval. Second, we uncover a complementary vulnerability in the fine-tuning stage: even benign training data can degrade safety. Building on this phenomenon, we develop a stronger red-teaming attack by identifying the most safety-degrading outliers within benign datasets. We introduce Self-Inf-N, an outlier-detection method that selects a small set of influential samples. Fine-tuning on as few as 100 selected outliers can severely compromise safety alignment across seven mainstream LLMs, with high transferability and poor coverage by existing mitigations. Third, motivated by these failures, we develop reasoning-aware alignment defenses. Using causal intervention, we show that many jailbreak vulnerabilities stem from shallow alignment behaviors that refuse without deeply understanding harmful intent. We address this by constructing a new Chain-of-Thought fine-tuning dataset spanning utility and safety-critical prompts to encourage principled, reasoned refusals. Finally, we propose Alignment-Weighted DPO, which assigns different preference weights to reasoning versus response segments, enabling targeted updates that improve jailbreak robustness while preserving utility. Together, these studies provide a unified framework for diagnosing and mitigating trustworthiness failures across RAG, fine-tuning, offering actionable guidance for deploying safer and fairer LLM systems.", "source_url": "https:\/\/openalex.org\/W7151325055", "pdf_url": "", "doi": "10.18130\/3vqg-ng96", "handle": "", "repository_name": "Libra", "source_record_id": "W7151325055", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:54", "updated_at": "2026-11-15 16:52:54" }, { "id": 944, "title": "Stance detection enhanced by large language models", "title_ar": "", "title_en": "Stance detection enhanced by large language models", "author": "Nikesh Gyawali", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Kansas State University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; nlp; medical-ai; responsible-ai; self-semi-supervised; llm; sentiment-analysis", "keywords": "", "abstract": "Stance detection, a subfield of opinion mining, is an important task in Natural Language Processing (NLP) that involves determining the attitude or position expressed by an author of a text towards a particular topic or claim, generally referred to as target.More specifically, the task involves automatically determining if an author of a text is In-Favor (Positive), Against (Negative), or Neutral towards a given target.Accurate stance detection is crucial for understanding complex social issues, guiding policy decisions, and shaping effective interventions.While the field of NLP has seen significant progress, capturing nuanced stances from diverse and often ambiguous data remains a challenge.Recent developments in Large Language Models (LLMs) have transformed the field by offering unprecedented capabilities in interpreting subtle linguistic cues and context-dependent meanings.This dissertation advances the field of stance detection by utilizing Large Language Models (LLMs) to improve methodologies across multiple critical and socially impactful domains.In this dissertation, we first explore LLM-enhanced approaches to stance detection in the socially controversial and polarizing topics of gun regulation and vaccines.We introduce the GunStance dataset, consisting of social media posts from X (formerly Twitter) posted by X users after major mass shooting events in the United States.By integrating labeled and unlabeled posts, this dataset allows comprehensive exploration of a semi-supervised learning framework in the context of stance detection.We propose a novel hybrid model that combines semi-supervised techniques with LLMs and show that our approach significantly outperforms traditional stance detection approaches.Furthermore, we assemble a large dataset of social media posts from X, capturing the vaccine discourse over a decade.The dataset includes seven years before COVID-19, as well as three COVID-19 years.Leveraging LLMs, social cognition theories and emotional dynamics, we analyze the vaccine dataset to capture the evolving public attitudes towards vaccines before and during the COVID-19 pandemic.Our study reveals increasing polarization and heightened emotional engagement, with a notable rise in vaccine skepticism amid the global health crisis.Expanding into the less controversial but important financial domain, we construct a financial stance detection corpus from annual 10-K reports filed to U.S. Securities and Exchange Commission (SEC) and earnings call transcripts (ECT) by extracting short text fragments relevant to key financial metrics, such as debt, earnings per share (EPS), and sales, and annotating them using LLM-driven methodologies with strict human validation.This financial stance detection corpus facilitates extensive evaluation of LLMs' ability to detect subtle stances towards financial metrics, a task that requires complex reasoning.Our findings demonstrate the effectiveness of LLMs in performing accurate stance detection without extensive labeled data, showcasing their potential for real-world financial analysis applications.Building upon these insights, we also introduce the Modular Prompt Optimization for Stance Detection (MoPrO-SD) framework.This framework utilizes the prompt optimization capabilities of LLMs by breaking down the complex stance detection prompt into modular, optimizable components.Each module is iteratively refined using LLMs as prompt optimizers, leading to an improved prompt that outperforms human-crafted prompts on several stance detection benchmarks.Collectively, this dissertation advances the field of stance detection by providing comprehensive evidence on the use of LLMs to enhance the performance, adaptability, and efficiency of stance detection methodologies across social media posts and financial documents, offering an analytical and scalable framework for informed and nuanced decision-making in an increasingly digital and interconnected world. Stance detection enhanced by large language models by", "source_url": "https:\/\/openalex.org\/W7218185954", "pdf_url": "https:\/\/krex.k-state.edu\/bitstreams\/ffe8f4b0-48a3-4b3c-8de7-53c0428e8d47\/download", "doi": "", "handle": "", "repository_name": "K-State Research Exchange (Kansas State University)", "source_record_id": "W7218185954", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:55", "updated_at": "2026-11-15 16:52:55" }, { "id": 952, "title": "Foundation models and large language models for autonomous driving", "title_ar": "", "title_en": "Foundation models and large language models for autonomous driving", "author": "Xiangrui Kong", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; nlp; generative-ai; robotics; responsible-ai; ai-applications; predictive-analytics; multimodal; llm; xai; ai-smart-cities", "keywords": "", "abstract": "The deployment of autonomous driving systems (ADS) is advancing globally, yet long-tail scenarios remain a major challenge. This thesis explores integrating foundation models (FMs) and large language models (LLMs) into ADS to enhance robustness, interpretability, and security. A multi-agent LLM framework is proposed for safe vehicle–LLM interaction, alongside an intent-aware pedestrian reasoning system. A generative framework combining LLMs, diffusion, and state-space models enables fine-grained 3D scene completion and traffic reporting. Further, a multimodal disengagement prediction model and secure path-planning architecture are developed. Collectively, this work provides a roadmap for embedding FMs into modular ADS pipelines.", "source_url": "https:\/\/openalex.org\/W7140680728", "pdf_url": "", "doi": "10.26182\/333t-y504", "handle": "", "repository_name": "UWA Profiles and Research Repository (UWA)", "source_record_id": "W7140680728", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:55", "updated_at": "2026-11-15 16:52:55" }, { "id": 961, "title": "Uncertainty in large language models", "title_ar": "", "title_en": "Uncertainty in large language models", "author": "Lukas Aichberger", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Johannes Kepler Universitat Linz", "university_source": "repository_host", "college": "", "department": "", "country": "Austria", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; computer-vision; nlp; predictive-analytics; multimodal; llm", "keywords": "", "abstract": "Large language models (LLMs) are increasingly serving as a primary interface for accessing information and interacting with digital systems. Nonetheless, LLMs can generate text that includes incorrect or unsupported claims while being fluent, coherent, and presented as factual. This creates a fundamental reliability issue: the generated text alone often does not reveal whether it can be trusted. Such unsupported factual claims are commonly referred to as LLM hallucinations. Detecting hallucinations is essential for deciding when a generated text can be trusted and when the system should instead abstain from answering, retrieve further evidence, or defer to a human. A principled approach to hallucination detection is uncertainty quantification, as elevated uncertainty is associated with a higher risk of hallucinations. However, estimating uncertainty in LLMs is challenging because uncertainty depends on the model’s predictive distribution over possible output sequences. Since LLMs stochastically generate output sequences token by token, this predictive distribution is high-dimensional and generally intractable to compute exactly. Thus, uncertainty estimation must rely on tractable approximations of this high-dimensional distribution. Developing effective approximations is a key enabler for the reliable deployment of LLM-based systems. In this cumulative thesis, we address the challenges posed by high-dimensional distributions to advance uncertainty estimation. We first establish foundational principles in the context of classical classification: our method QUAM demonstrates that uncertainty estimation benefits from targeted exploration, explicitly searching for samples that most influence the estimate. We then translate this insight to LLMs through SDLG, a method that improves uncertainty estimation by sampling output sequences that are most informative about the underlying distribution. Furthermore, we show that competitive uncertainty estimates can also be obtained without sampling multiple output sequences by leveraging a single-sequence measure: the negative log-likelihood of the most likely output sequence, approximated through our method G-NLL. This provides a highly efficient and practical alternative to sampling-based uncertainty estimation in LLMs. While this thesis primarily focuses on the internal reliability of LLMs through uncertainty estimation, it also investigates the external reliability of LLM-based systems deployed as multimodal agents that observe and act within their environment. To this end, we introduce Malicious Image Patches (MIPs), demonstrating that multimodal agents can be compromised into executing malicious actions when such patches appear in their visual observations. By exposing this vulnerability, this work highlights the need for safeguards against external failure modes and broadens the thesis from the internal to the external reliability of systems built around LLMs. Overall, the core contributions of this thesis advance uncertainty estimation in LLMs and thereby improve uncertainty-based hallucination detection. Complemented by insights into malicious failure modes, these contributions provide a principled foundation for improving the internal and external reliability of LLM-based systems.", "source_url": "https:\/\/openalex.org\/W7169636052", "pdf_url": "", "doi": "", "handle": "", "repository_name": "University Library Linz repository (Johannes Kepler Universitat Linz)", "source_record_id": "W7169636052", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:55", "updated_at": "2026-11-15 16:52:55" }, { "id": 964, "title": "Transcreation Applied to Large Language Models", "title_ar": "", "title_en": "Transcreation Applied to Large Language Models", "author": "Vieira, Inês Mendes", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Portugal", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "nlp; llm", "keywords": "", "abstract": "The aim of this internship report is to analyze the potential of transcreation through prompt engineering in Large Language Models in the particular cases of customer support and marketing. This report explores prompt engineering as an act of transcreation, delving into how it influences the output generated by these models. In a first instance, this report seeks to evaluate the potential of transcreation through prompt engineering in customer support based on translation error data obtained from annotations. In a second instance, it seeks to explore how prompt engineering improves upon outputs when considering transcreation in marketing. Finally, it considers the practical applications of the findings obtained from both instances of research, customer support and marketing, as well as how they may be harnessed into further future development, with results enforcing the idea that this is a feasible area of research with promising applications", "source_url": "https:\/\/openalex.org\/W7112047225", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT)", "source_record_id": "W7112047225", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:56", "updated_at": "2026-11-15 16:52:56" }, { "id": 970, "title": "Cognitive Behaviours of Large Language Models", "title_ar": "", "title_en": "Cognitive Behaviours of Large Language Models", "author": "Bryan Wilie", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; computer-vision; nlp; reinforcement-learning; transfer-learning; llm", "keywords": "", "abstract": "Large language models (LLMs) have reshaped natural language processing, demonstrating remarkable reasoning across diverse tasks. Their rapid real-world integration has fostered growing reliance on their perceived cognitive abilities, with users expecting models to adapt to evolving information, perform consistently across languages, and handle precise computations. However, a central question remains: to what extent do LLMs exhibit consistent cognitive behaviors under dynamic, diverse real-world conditions? This thesis explores LLM use cases in practical settings requiring adaptability, inclusivity across multilingual users, and computational precision. First, we study LLMs’ cognitive limitations in revising inferences when presented with new, contradictory, or clarifying information. Existing benchmarks assume complete context, neglecting information’s dynamic nature. We introduce a sequential evaluation framework simulating scenarios requiring belief revision, revealing that strong LLMs often cannot revise prior beliefs. We identify a fundamental trade-off: models adept at updating underperform when no revision is necessary, and vice versa—exposing key tensions in model adaptability. Second, we analyze whether LLM cognitive behaviors remain consistent across languages. Despite training on large multilingual corpora, LLM capabilities remain uneven across languages. Analyzing internal representations, we uncover inconsistent interlingual alignment and show that monolingual fine-tuning further disrupts this alignment. To mitigate this, we propose a selective freezing strategy during fine-tuning that preserves interlingual representation alignment and improves cross-lingual generalization. Third, we address LLM limitations in tasks requiring precise computation, particularly in LLM-based judges for domains demanding mathematical exactness. Programmatic tools retain advantages over purely language-based reasoning. We develop a multi-turn reinforcement learning framework enabling LLM judges to selectively invoke an external Python interpreter as part of their reasoning process. This framework enables judges to learn beneficial tool invocation strategies, delegating sub-problems requiring exact computation to code execution. Judges integrate execution outputs into their chain-of-thought reasoning, mitigating errors inherent to text-based computation and enhancing judgment accuracy.", "source_url": "https:\/\/openalex.org\/W7135076034", "pdf_url": "", "doi": "10.14711\/thesis-hdl169569", "handle": "", "repository_name": "", "source_record_id": "W7135076034", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:56", "updated_at": "2026-11-15 16:52:56" }, { "id": 972, "title": "Quality Assurance in Large Language Models", "title_ar": "", "title_en": "Quality Assurance in Large Language Models", "author": "Song, Da", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Alberta", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; nlp; medical-ai; responsible-ai; data-mining; predictive-analytics; llm; xai; fairness", "keywords": "", "abstract": "Large Language Models (LLMs) have demonstrated impressive capabilities across a wide range of tasks in natural language processing (NLP) and software engineering (SE). However, their increasing adoption has raised critical concerns about trustworthiness, including robustness under distribution shifts, vulnerability to hallucinations, and misalignment between confidence and accuracy. These issues are exacerbated by the complex and probabilistic nature of LLMs, which makes it difficult for developers to observe, interpret, and address their failures. To bridge this gap, this thesis presents a series of tools and frameworks that enable a more transparent and actionable understanding of LLM trustworthiness. Together, these works follow a methodological arc that moves from interactive exploration to internal structure modeling and finally to external behavioral correction. The first contribution, DeepLens, introduces a human-centered system that supports interactive diagnosis of out-of-distribution (OOD) data in large text corpora. By integrating clustering-based analysis and neuron-level highlighting, {\\deeplens} empowers developers to detect, categorize, and interpret OOD issues that threaten model reliability. Building on this foundation, the second contribution, LUNA, proposes a white-box framework that models LLM behavior using abstract probabilistic structures. By mapping hidden states to interpretable semantic concepts, LUNA enables structured evaluation of trustworthiness dimensions such as hallucination and adversarial robustness, providing insights grounded in both semantic and model-level perspectives. The final contribution addresses the limitations of white-box methods in black-box scenarios. It introduces a confidence calibration framework that quantifies and corrects model confidence based on semantic-preserving behavioral variations. By leveraging agreement ratios, entropy, and stability features from multiple LLM outputs, this method learns a regression-based correction factor to improve alignment between model confidence and factual correctness, without requiring model internals. Together, these three works provide a practical toolchain to enhance the transparency and trustworthiness of LLMs, spanning interactive user interfaces, interpretable model-based analysis, and scalable confidence calibration.", "source_url": "https:\/\/openalex.org\/W7105991970", "pdf_url": "", "doi": "10.7939\/83296", "handle": "", "repository_name": "ERA: Education and Research Archive (University of Alberta)", "source_record_id": "W7105991970", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:56", "updated_at": "2026-11-15 16:52:56" }, { "id": 977, "title": "Large Language Model Evaluation", "title_ar": "", "title_en": "Large Language Model Evaluation", "author": "優介 坂井", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Japan", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "nlp; llm", "keywords": "", "abstract": "Large language models (LLMs) have become foundational resources in the natural language processing field.The capabilities of LLMs in inference, natural language understanding (NLU), and their pre-trained knowledge are widely utilized in various applications.To fully leverage these capabilities, it is essential to properly evaluate them.However, it is very challenging to evaluate LLMs accurately.For example, when evaluating an LLM's inference capabilities to infer unseen facts from seen ones, it may rely on memorized knowledge from pre-training rather than pure inference, due to the pre-training datasets being curated from web data, which can lead to contamination with related knowledge.Therefore, traditional evaluation frameworks for inference capabilities struggle to capture the pure inference capability of LLMs, independent of their pre-trained knowledge.Furthermore, when evaluating the NLU capabilities of LLMs, we should consider the stability in outputs for various input prompts.Finally, constructing evaluation datasets involves significant costs of financial, time, and human labor.Especially when focusing on multilingual datasets, there is a limitation in the availability of annotators, so most datasets are machine-translated from English.However, due to cultural differences and noise introduced by translation, these evaluations are unsuitable for language-specific evaluation.This disserta-", "source_url": "https:\/\/openalex.org\/W7143759745", "pdf_url": "https:\/\/naist.repo.nii.ac.jp\/record\/2000625\/files\/R019175.pdf", "doi": "10.34413\/dr.02100", "handle": "", "repository_name": "Institutional Repositories DataBase (IRDB)", "source_record_id": "W7143759745", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:56", "updated_at": "2026-11-15 16:52:56" }, { "id": 1011, "title": "Reconciling with Noise in Machine Learning for Healthcare", "title_ar": "", "title_en": "Reconciling with Noise in Machine Learning for Healthcare", "author": "Sujay Nagaraj", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Toronto", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "medical-ai", "research_subdomain": "fairness", "categories": "machine-learning; medical-ai; responsible-ai; ai-applications; predictive-analytics; fairness; ai-industry", "keywords": "", "abstract": "This thesis explores the challenges presented by label noise in machine learning (ML) for health. Label noise is unavoidable in real-world data; it can arise from variability in measurement, human annotation errors, and aleatoric uncertainty in the ground truth. Ignoring label noise when training classifiers with standard risk-minimization algorithms can result in faulty models that predict noise instead of signal; this is particularly problematic in healthcare where predictions can representmission-critical decisions. This thesis addresses the problem of label noise from several perspectives: exploring how and why label noise arises in health data, proposing robust algorithms to handle label noise, and studying the impact of label noise on individualized predictions while proposing methods for safer inference. Through a real-world study of stress detection in sensor data, the thesis first explores the significant inter- and intra-individual variability in label representations, undermining the reliability of classifiers trained on these labels. Motivated by this, the thesis introduces the concept of Temporal Label Noise, where noise levels vary over time, and proposes robust algorithms to handle this type of noise. It then examines the consequences of label noise on individual-level predictions, a gap in existing label noise research, while providing tools to mitigate the potential harm in high-stakes healthcare predictions. Finally, the work delves into a clinical case study in the Critical Care Unit, demonstrating how noisy sensor data can be leveraged to generate higher-quality annotations and improve patient safety initiatives. By identifying clinical challenges and proposing novel algorithmic solutions, this thesis intends to help advance the development of safe, robust, and trustworthy machine learning for healthcare.", "source_url": "https:\/\/openalex.org\/W7132904433", "pdf_url": "https:\/\/utoronto.scholaris.ca\/bitstreams\/9822ef19-c180-4bf4-96cb-b050f266a01b\/download", "doi": "", "handle": "", "repository_name": "TSpace (University of Toronto)", "source_record_id": "W7132904433", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1016, "title": "Machine learning for healthcare with a focus on the early diagnosis of epilepsy and brain tumor detection", "title_ar": "", "title_en": "Machine learning for healthcare with a focus on the early diagnosis of epilepsy and brain tumor detection", "author": "Diyuan Lu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Goethe University Frankfurt", "university_source": "repository_host", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; predictive-analytics; rnn; medical-imaging", "keywords": "", "abstract": "Machine learning (ML) techniques have evolved rapidly in recent years and have shown impressive capabilities in feature extraction, pattern recognition, and causal inference. There has been an increasing attention to applying ML to medical applications, such as medical diagnosis, drug discovery, personalized medicine, and numerous other medical problems. ML-based methods have the advantage of processing vast amounts of data. With an ever increasing amount of medical data collection and large, inter-subject variability in the medical data, automated data processing pipelines are very much desirable since it is laborious, expensive, and error-prone to rely solely on human processing. ML methods have the potential to uncover interesting patterns, unravel correlations between complex features, learn patient-specific representations, and make accurate predictions. Motivated by these promising aspects, in this thesis, I present studies where I have implemented deep neural networks for the early diagnosis of epilepsy based on electroencephalography (EEG) data and brain tumor detection based on magnetic resonance spectroscopy (MRS) data. In the project for early diagnosis of epilepsy, we are dealing with one of the most common neurological disorders, epilepsy, which is characterized by recurrent unprovoked seizures. It can be triggered by a variety of initial brain injuries and manifests itself after a time window which is called the latent period. During this period, a cascade of structural and functional brain alterations takes place leading to an increased seizure susceptibility. The development and extension of brain tissue capable of generating spontaneous seizures is defined as epileptogenesis (EPG). Detecting the presence of EPG provides a precious opportunity for targeted early medical interventions and, thus, can slow down or even halt the disease progression. In order to study brain signals in this latent window, animal epilepsy models are used to provide valuable data as it is extremely difficult to obtain this data from human patients. The aim of this study is to discover biomarkers of EPG using animal models and then to find the equivalent and counterparts in human patients' data. However, the EEG features for EPG are not well-understood and there is not a sufficiently large amount of annotated data for ML-based algorithms. To approach this problem, firstly, I utilized the timestamp information of the recorded EEG from an animal epilepsy model where epilepsy is induced by an electrical stimulation. The timestamp serves as a form of weak supervision, i.e., before and after the stimulation. Secondly, I implemented a deep residual neural network and trained it with a binary classification task to distinguish the EEG signals from these two phases. After obtaining a high discriminative ability on the binary classification task, I proposed to divide further the time span after the stimulation for a three-class classification, aiming to detect possible stages of the progression of the latent EPG phase. I have shown that the model can distinguish EEG signals at different stages of EPG with high accuracy and generalization ability. I have also demonstrated that some of the learned features from the network are clinically relevant. In the task of detecting brain tumors based on MRS data, I first proposed to apply a deep neural network on the MRS data collected from over 400 patients for a binary classification task. To combat the challenge of noisy labeling, I developed a distillation step to filter out relatively ``cleanly'' labeled samples. A mixing-based data augmentation method was also implemented to expand the size of the training set. All the experiments were designed to be conducted with a leave-patient-out scheme to ensure the generalization ability of the model. Averaged across all leave-patient-out cross-validation sets, the proposed method performed on par with human neuroradiologists, while outperforming other baseline methods. I have demonstrated the distillation effect on the MNIST data set with manually-introduced label noise as well as providing visualization of the input influences on the final classification through a class activation map method. Moreover, I have proposed to aggregate information at the subject level, which could provide more information and insights. This is inspired by the concept of multiple instance learning, where instance-level labels are not required and which is more tolerant to noisy labeling. I have proposed to generate data bags consisting of instances from each patient and also proposed two modules to ensure permutation invariance, i.e., an attention module and a pooling module.", "source_url": "https:\/\/openalex.org\/W4229378113", "pdf_url": "http:\/\/publikationen.ub.uni-frankfurt.de\/files\/67966\/Diyuan_PhD_Thesis-pdfA.pdf", "doi": "10.21248\/gups.67966", "handle": "", "repository_name": "Publication Server of Goethe University Frankfurt am Main (Goethe University Frankfurt)", "source_record_id": "W4229378113", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1021, "title": "Optimization and Machine Learning Methods for Medical and Healthcare Applications", "title_ar": "", "title_en": "Optimization and Machine Learning Methods for Medical and Healthcare Applications", "author": "Cao Xiao", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Washington", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2016, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; computer-vision; medical-ai; predictive-analytics; medical-imaging", "keywords": "", "abstract": "The increasing amounts of data being gathered in healthcare and medical systems and the convergence of different domains are leading medical and healthcare research to a new direction of precision and personalized medicine. The trends bring a unique opportunity and good promise to solving various critical tasks in medical and healthcare research. However, such promise heavily relies on whether we could find useful patterns to characterize the target problems, discover informative mechanisms underlying the noisy and fragmented factual information, as well as transform these knowledge into intelligent decision making. Previously many efforts have been made via various approaches such as machine learning, optimization, statistical analysis, mathematical modeling, biomedical informatics, etc. This thesis will extend along the lines of machine learning and optimization methods, and to build novel models for solving the medical or healthcare challenges. Particularly, the thesis focus on the following topic domains, including medical image or recording based disease differentiation and diagnosis, image guided precision surgery, disease progression modeling, and personalized health behavior recommendation. To summarize, the thesis mainly includes five models that are fit for a wide varieties of healthcare and medical applications. The first model is an integrated feature ranking and selection framework that is capable of selecting a sparse model while preserving the most informative features. The framework combines information theoretic criteria and the least absolute shrinkage and selection operator (lasso) method into a two-step feature selection process. It can be applied to biomarker selection problems when the number of subjects is small comparing with number of candidate biomarkers. The second is a structure learning model capable of achieving personalized identification of surgery insertion location. I introduce a method to craft novel patient-specific features from their medical images. In addition, I propose a supervised structure learning and prediction model with special inter-dimensional and response structure regularization terms to capture spatial relations of features and responses. The third is a systematic framework that unifies dynamic modeling, sparse learning, dictionary learning, and matrix completion to translate users’ behavioral data into deeply personalized health planning. The framework is fit for longitudinal user behavior data, and can potentially be a backend algorithm for mobile health (mHealth) technologies. The fourth model is an optimal expert knowledge elicitation strategy for identifying pairwise Bayesian network structure from observational data. It combines observational data and expert knowledge and iteratively elicit new expert knowledge that is optimally matched to the observational data to maximally reduce the uncertainty in the structure identification. The strategy can be applied to leverage expert knowledge in learning causal relation from data when the observational data is limited. In the thesis we use it to learn the underlying progression mechanism of Alzheimer’s disease. Last, the thesis also presents a decomposed version of the k nearest neighbor (DKNN) method for the classification of an unseen object based on the distances to the centroids of the k nearest neighbors. The DKNN has a training process to learn the local-optima-free distance metric by solving a convex optimization problem. The optimization problem not only learns a metric that minimizes classification errors and maximizes the margin between intra-class and inter-class distance. In addition, it also selects important features and removes irrelevant features when L1 regularization is incorporated in the optimization model. The DKNN algorithm is designed as a general method, but could be used in identifying discriminating features for a particular disease (binary cases) or different patient cohorts (multi-class cases). To demonstrate the utility of the proposed machine learning and optimization models, numerical studies are performed using simulation, public and\/or real data. Real data including magnetic resonance imaging (MRI) data, electroencephalogram (EEG) recordings, positron emission tomography (PET) data, data from wearable devices, computed tomography (CT) scan, etc. Through extensive experiments and analysis, the proposed approaches outperform baseline methods, and demonstrate their utility and efficacy in medical and healthcare applications.", "source_url": "https:\/\/openalex.org\/W2597649913", "pdf_url": "", "doi": "", "handle": "", "repository_name": "ResearchWorks at the University of Washington (University of Washington)", "source_record_id": "W2597649913", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1023, "title": "Epilepsy diagnosis from scalp EEG : a machine learning approach", "title_ar": "", "title_en": "Epilepsy diagnosis from scalp EEG : a machine learning approach", "author": "Thomas, John", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Nanyang Technological University", "university_source": "repository_host", "college": "", "department": "", "country": "Singapore", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "medical-ai", "research_subdomain": "rnn", "categories": "machine-learning; deep-learning; medical-ai; rnn", "keywords": "", "abstract": "Epilepsy refers to a group of chronic brain disorders characterized by recurrent unprovoked seizures. Interictal Epileptiform Discharges (IEDs) are distinctive biomarkers of epilepsy. According to the International Federation of Societies for Electroencephalography and Clinical Neurophysiology (IFSECN), IEDs are defined as transients distinguishable from background activity, with a characteristic spiky morphology, typically, but neither exclusively nor invariably, found in interictal electroencephalograms (EEGs) of people with epilepsy. Approximately 65 million people are affected worldwide. The early diagnosis of epilepsy is based on the IEDs that appear in the EEG. The EEGs are non-invasive, inexpensive, digitized recordings of brain activity from the scalp. However, there is a lack of experts who can analyze EEGs; moreover, manual interpretation EEG is time-consuming. Moreover, IEDs exhibit huge variation across patients. Besides, a standard definition of interictal epileptiform events is yet to be formulated. This makes the IED detection based diagnosis of epilepsy, tedious, and expert-centered. Experts have a significant disagreement over IED annotation. Consequently, an auvii", "source_url": "https:\/\/openalex.org\/W2972587271", "pdf_url": "https:\/\/dr.ntu.edu.sg\/bitstream\/10356\/93579\/1\/THOMAS_JOHN_THESIS.pdf", "doi": "10.32657\/10356\/93579", "handle": "", "repository_name": "DR-NTU (Nanyang Technological University)", "source_record_id": "W2972587271", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1026, "title": "Machine learning for childhood pneumonia diagnosis", "title_ar": "", "title_en": "Machine learning for childhood pneumonia diagnosis", "author": "Elina Naydenova", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Oxford", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; medical-ai; ai-applications; medical-imaging; ai-smart-cities", "keywords": "", "abstract": "Pneumonia is the number one killer of children under the age of 5, causing more deaths than malaria, tuberculosis and HIV\/AIDS combined. In 2015, over 920,000 children died of pneumonia and more than 95% of the incidence and 99% of subsequent mortality occurred in developing countries. Current gold standard diagnostic assessment of childhood pneumonia relies on the use of advanced tools (such as X-rays and blood culture) by a clinical expert who assesses and interprets a combination of clinical measurements. However, the shortage of clinical expertise and equipment in low-resource settings delay diagnosis, increasing the risk of mortality. This thesis investigates the role that machine learning could play in improving diagnosis of childhood pneumonia in low-resource settings. Point-of-care devices such as a digital stethoscope and a pulse oximeter have become more affordable and hence abundant across the globe. The adoption of such devices, particularly in community settings, presents an opportunity to leverage information captured through their signals to provide early diagnosis for conditions such as childhood pneumonia. Hence, this thesis proposes a machine learning framework for the design of diagnostic models, built on a parsimonious set of symptoms which can be captured in a point-of-care setting, to deliver accurate and reproducible diagnosis of pneumonia. The approach is evaluated on three different datasets (with accuracy ranging between 81.8 – 97.9%), consistently outperforming diagnosis with the current gold standard guidelines for community diagnosis by the World Health Organisation. A clinical study in Mumbai, India, is designed to capture a realistic dataset in a poor urban community and a resource-constrained hospital, where a mobile health toolkit including three point-of-care devices (a digital stethoscope, a pulse oximeter and a thermometer) is built to enable minimally trained users to capture essential health information. Diagnostic models developed on this rich dataset are seen to perform well when validated internally as well as generalise to other datasets from different geographies. Finally, a set of tools for automated analysis of lung signals is proposed and validated on two big datasets, one from Peru and one from India. Combined with vital signs, information derived through this automated lung signal analysis was seen to outperform diagnosis via expert clinical annotation, delivering 81.3% sensitivity and 84.5% specificity.", "source_url": "https:\/\/openalex.org\/W2963713112", "pdf_url": "https:\/\/ora.ox.ac.uk\/objects\/uuid:1fb64eee-10aa-43cf-84c0-e46dd49202a9\/files\/m7bcc3a360014e80d2fa86f9d554502ce", "doi": "10.5287\/ora-vq6xvo91j", "handle": "", "repository_name": "Oxford University Research Archive (ORA) (University of Oxford)", "source_record_id": "W2963713112", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1027, "title": "Towards Practical Applications of Machine Learning in Healthcare with Federated Learning", "title_ar": "", "title_en": "Towards Practical Applications of Machine Learning in Healthcare with Federated Learning", "author": "Zhong, Aoxiao", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Harvard University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; medical-ai; responsible-ai; predictive-analytics; medical-imaging; federated-learning; fairness", "keywords": "", "abstract": "Federated Learning (FL) has emerged as a significant tool in healthcare machine learning, enabling institutions to collaboratively train models while maintaining data privacy. This dissertation describes the implementation of a real-world healthcare FL project and addresses the challenge of domain shift for more effective model deployment. We begin by detailing a practical application of FL during the SARS-COV-2 pandemic. Twenty institutions collaborated on a healthcare FL study to develop the \"EXAM\" (EMR CXR AI Model), which predicts future oxygen requirements for symptomatic patients using vital signs, laboratory data, and chest x-rays. EXAM achieved an Area Under the Curve (AUC) of over 0.92, marking a 16\\% improvement and a 38\\% increase in generalizability over local models. This project demonstrated FL's ability to enable rapid scientific collaboration without data exchange, producing a model that generalized across heterogeneous, unharmonized datasets and provided the healthcare community with a validated tool to combat COVID-19. Next, we address a specialized non-iid FL challenge termed \\emph{Domain-mixed FL}, where each client's data is assumed to be a mixture of several predefined domains. We propose a novel method, FedDAR, which learns a shared domain representation and personalized prediction models in a decoupled manner. Theoretical proofs show that FedDAR achieves linear convergence in simplified settings, and extensive empirical studies on both synthetic and real-world datasets demonstrate its superiority over existing FL methods. Finally, we explore the multi-dimensional domain shift problem prevalent in healthcare ML applications. We introduce a novel strategy using an ensemble of mixtures of experts (EMoE), each expert tailored to adapt to shifts along different dimensions. This approach is designed to be versatile and robust, suitable for both centralized and federated learning settings. Rigorous testing on various real-world datasets has shown that our method outperforms contemporary domain generalization and personalized federated learning approaches, effectively managing the complexities of multi-dimensional domain shifts.", "source_url": "https:\/\/openalex.org\/W7113010396", "pdf_url": "https:\/\/dash.harvard.edu\/bitstreams\/ac3d7a71-38eb-4185-a983-9d0d8a225bf3\/download", "doi": "", "handle": "", "repository_name": "Digital Access to Scholarship at Harvard (DASH) (Harvard University)", "source_record_id": "W7113010396", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1030, "title": "Towards robust real-world historical handwriting recognition", "title_ar": "", "title_en": "Towards robust real-world historical handwriting recognition", "author": "Mahya Ameryan", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Groningen \/ Centre for Information Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Netherlands", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "computer-vision", "research_subdomain": "image-classification", "categories": "computer-vision; ai-applications; image-classification; ai-education", "keywords": "", "abstract": "Historical manuscripts embody important witnesses of cultural heritage.In recent years, libraries all over the world have scanned voluminous document collections and often made the scanned page images available online.However, especially for handwritten manuscripts, having a text image is not the same as having access to digital text content.Wouldn't it be great if you type just in a few words on such a website, and then you get into the relevant section of a historical document?Given the current development of advanced artificial-intelligence algorithms and the availability of fast, powerful computers, one would expect that this problem has been solved.However, a perfect translation of the pixel image to a correct Unicode text version of that document is it not generally possible [1], [2].Even modern OCR tools developed by Google, like Tesseract [3], [4] and Google document AI, as well as the well-known Google Book collection of scanned, machine-printed books cannot provide accurate text-recognition results which are acceptable in the detail [5]-[7].This lack of accuracy is especially problematic for documents that are intended for scholarly research, where the demand for accurate transcriptions is high.Even advanced machine-learning methods struggle with the recognition of machine-printed documents when the machine font shapes are complex, e.g., in 'Gothic' [8].Given this state of affairs for the case of machine-printed books one may appreciate that handwriting, with its variation and inherent sloppiness, poses many additional problems.Handwriting recognition by computers therefore is a challenging dynamic field of research in the artificial intelligence domain [9]- [11].The extraordinary performance (99.91% [12]) on the well-known MNIST benchmark data set [13] of handwritten digits may give the impression that artificial-intelligence research has solved the problem of recognizing human handwriting.As we will also explain in the section 1.1, this is not yet the case.Even well-curated 'academic' benchmarks for word recognition such as 'IAM' [14], 'RIMES' [15] and 'GW' [16] yield word-recognition accuracy performances of 95-98%.This means: a word error in every 20-50 words.Such performances are not acceptable for scholarly research in the humanities.For realistic,", "source_url": "https:\/\/openalex.org\/W4389087222", "pdf_url": "https:\/\/research.rug.nl\/files\/837278989\/Complete_thesis.pdf", "doi": "10.33612\/diss.837278987", "handle": "", "repository_name": "University of Groningen research database (University of Groningen \/ Centre for Information Technology)", "source_record_id": "W4389087222", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1032, "title": "Deep Learning-Based Text Detection and Recognition", "title_ar": "", "title_en": "Deep Learning-Based Text Detection and Recognition", "author": "Qingqing Wang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Technology Sydney", "university_source": "repository_host", "college": "", "department": "", "country": "Australia", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; deep-learning; computer-vision; predictive-analytics; cnn; rnn; image-classification; object-detection; image-segmentation", "keywords": "", "abstract": "Texts play a critical role in our daily life. They are everywhere such as slogans on posters, licence plates on cars, etc., to transmit information and knowledge. With the popularity of mobile devices with cameras, more and more texts are collected, transmitted and stored as text images. Automatically reading texts from images is of high application potentials. Therefore, related researches have been attracting considerable attentions from the computer vision community. Scene texts and handwritten texts are the two most difficult texts to be automatically read because of the challenges posed by the complexity of backgrounds, the uncertainty of capturing conditions, the diversity of text appearances, touching characters and the variety of handwriting styles. Text detection, 𝒾.𝑒., localizing text areas from images, and text recognition, 𝒾.𝑒., transcribing located text areas into character sequences, are two key steps of robust text reading. In recent years, they have entered a deep learning era, where Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) play important roles. Here, we conduct researches on text detection and recognition based on CNN and LSTM, as presented below. 1. To improve the recall rate of small text areas in oriented text detection, we propose an Xception-based multi-ASPP-assembled scene text detector named DeepText. DeepText inserts multiple Atrous Spatial Pyramid Pooling (ASPP) modules into Xception after feature maps with different resolutions to retain richer information for small text areas, and introduces auxiliary connections and auxiliary losses to speed up convergence and boost the discrimination ability of lower encoder layers. 2. To address the issue that Mask R-CNN cannot fully leverage global information when performing predictions, we propose a scene text detector named GMask R-CNN, where a global mask module is designed to perform semantic segmentation by considering global information. 3. To tackle the problem that LSTM neglects the valuable spatial and structural information of 2-D text images, we propose two scene text recognisers named FACLSTM, which exploits convolution LSTM to directly perform sequential transcription in 2-D space, and ReELFA, which utilizes one-hot encoded locations to enhance features with pixels' spatial information. 4. To solve the problem that CNNs with fully connected layers are not suitable for sequential prediction tasks due to their requirements of fixed-size inputs\/outputs, we propose a CNN-based handwritten text recogniser CFSPP. CFSPP embeds a Spatial Pyramid Pooling-based intermediate layer between convolutional layers and fully connected layers to convert arbitrary-size feature maps into feature vectors with specific lengths.", "source_url": "https:\/\/openalex.org\/W3154876829", "pdf_url": "", "doi": "", "handle": "", "repository_name": "UTS ePRESS (University of Technology Sydney)", "source_record_id": "W3154876829", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1034, "title": "Deep Learning for Amharic Text-Image Recognition: Algorithm, Dataset and Application", "title_ar": "", "title_en": "Deep Learning for Amharic Text-Image Recognition: Algorithm, Dataset and Application", "author": "Birhanu Hailu Belay", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Kaiserslautern University of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; nlp; ai-applications; cnn; rnn; transformers; image-classification; machine-translation; ai-industry", "keywords": "", "abstract": "Optical Character Recognition (OCR) is one of the central problems in pattern recognition. Its applications have played a great role in the digitization of document images collected from het- erogeneous sources. Many of the well-known scripts have OCR systems with sufficiently high performance that enables OCR applications in industrial\/commercial settings. However, OCR sys- tems yield very-good results only on a narrow domain and very-specific use cases. Thus, it is still a challenging task, and there are other exotic languages with indigenous scripts, such as Amharic, for which no well-developed OCR systems exist. As many as 100 million people speak Amharic, and it is an official working language of Ethiopia. Amharic script contains about 317 different alphabets derived from 34 consonants with small changes. The change involves shortening or elongating one of its main legs or adding small diacritics to the right, left, top, or bottom of the consonant character. Such modifications lead the characters to have similar shapes and make the recognition task complex, but this is particularly interesting for charac- ter recognition research. So far, Amharic script recognition models are developed based on classical machine learning techniques, and they are very limited in addressing the issues for Amharic OCR. The motivation of this thesis is, therefore, to explore and tailor contemporary deep learning tech- niques for the OCR of Amharic. This thesis addresses the challenges in Amharic OCR through two main contributions. The first contribution is an algorithmic contribution in which we investigate deep learning approaches that suit the demand for Amharic OCR. The second is a technical contribution that comprises several works towards the OCR model development; thus, it introduces a new Amharic database consisting of collections of images annotated at a character and text-line level. It also presents a novel CNN- based framework designed by leveraging the grapheme of characters in Fidel-Gebeta (where Fidel- Gebeta consists of the full set of Amharic characters in matrix structure) and achieves 94.97% overall character recognition accuracy. In addition to character level methods, text-line level methods are also investigated and devel- oped based on sequence-to-sequence learning. These models avoid several of the pre-processing stages used in prior works by eliminating the need to segment individual characters. In this design, we use a stack of CNNs, before the Bi-LSTM layers and train from end-to-end. This model out- performs the LSTM-CTC based network, on average, by a CER of 3.75% with the ADOCR test set. Motivated by the success of attention, in addressing the problems’ of long sequences in Neural Machine Translation (NMT), we proposed a novel attention-based methodology by blending the attention mechanism into CTC objective function. This model performs far better than the existing techniques with a CER of 1.04% and 0.93% on printed and synthetic text-line images respectively. Finally, this thesis provides details on various tasks that have been performed for the development of Amharic OCR. As per our empirical analysis, the majority of the errors are due to poor annotation of the dataset. As future work, the methods proposed in this thesis should be further investigated and extended to deal with handwritten and historical Amharic documents.", "source_url": "https:\/\/openalex.org\/W3163416829", "pdf_url": "https:\/\/kluedo.ub.uni-kl.de\/frontdoor\/index\/index\/docId\/6291", "doi": "10.26204\/kluedo\/6291", "handle": "", "repository_name": "Publication Server of Kaiserslautern University of Technology (Kaiserslautern University of Technology)", "source_record_id": "W3163416829", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1036, "title": "Application of automatic speech recognition technologies to singing", "title_ar": "", "title_en": "Application of automatic speech recognition technologies to singing", "author": "Anna Kruspe", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "nlp; speech-recognition", "keywords": "filler HMMs for keyword spotting, an i-vector approach for language identification", "abstract": "The research field of Music Information Retrieval is concerned with the automatic analysis of musical characteristics. One aspect that has not received much attention so far is the automatic analysis of sung lyrics. On the other hand, the field of Automatic Speech Recognition has produced many methods for the automatic analysis of speech, but those have rarely been employed for singing. This thesis analyzes the feasibility of applying various speech recognition methods to singing, and suggests adaptations. In addition, the routes to practical applications for these systems are described. Five tasks are considered: Phoneme recognition, language identification, keyword spotting, lyrics-to-audio alignment, and retrieval of lyrics from sung queries. The main bottleneck in almost all of these tasks lies in the recognition of phonemes from sung audio. Conventional models trained on speech do not perform well when applied to singing. Training models on singing is difficult due to a lack of annotated data. This thesis offers two approaches for generating such data sets. For the first one, speech recordings are made more “song-like”. In the second approach, textual lyrics are automatically aligned to an existing singing data set. In both cases, these new data sets are then used for training new acoustic models, offering considerable improvements over models trained on speech. Building on these improved acoustic models, speech recognition algorithms for the individual tasks were adapted to singing by either improving their robustness to the differing characteristics of singing, or by exploiting the specific features of singing performances. Examples of improving robustness include the use of keyword-filler HMMs for keyword spotting, an i-vector approach for language identification, and a method for alignment and lyrics retrieval that allows highly varying durations. Features of singing are utilized in various ways: In an approach for language identification that is well-suited for long recordings; in a method for keyword spotting based on phoneme durations in singing; and in an algorithm for alignment and retrieval that exploits known phoneme confusions in singing.", "source_url": "https:\/\/openalex.org\/W2945891342", "pdf_url": "https:\/\/www.db-thueringen.de\/servlets\/MCRFileNodeServlet\/dbt_derivate_00041256\/ilm1-2018000226.pdf", "doi": "", "handle": "", "repository_name": "Common Library Network (Der Gemeinsame Bibliotheksverbund)", "source_record_id": "W2945891342", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1041, "title": "Deep learning with synthetic, temporal, and adversarial supervision", "title_ar": "", "title_en": "Deep learning with synthetic, temporal, and adversarial supervision", "author": "Ankush Gupta", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Oxford", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; deep-learning; computer-vision; predictive-analytics; cnn; rnn; object-detection; video-analysis", "keywords": "", "abstract": "In this thesis we explore alternatives to manually annotated training examples for supervising the training of deep learning models. Specifically, we develop methods for learning under three different supervision paradigms, namely — (1) synthetic data, (2) temporal data, and (3) adversarial supervision for learning from unaligned examples. The dominant application domain of our work is text spotting, i.e. detection and recognition of text instances in images. We learn text localisation networks on synthetic data, and harness an adversarial discriminator for training text recognition networks using no paired training examples. Further, we exploit the changing pose of objects in temporal sequences (videos) to learn object landmark detectors. The unifying objective is to scale deep learning methods beyond manually annotated training data. We develop a large-scale, realistic synthetic scene text dataset. Armed with this large annotated dataset of scene images, we train a novel, fast fully-convolutional text detection network, and show excellent performance on real images. This generalisation from synthetic to real images, confirms the verisimilitude of our rendering process. The dataset, SynthText in the Wild, has been widely adapted by the research community, and has enabled the development of end-to-end text spotting models. While synthetic text can be readily generated, it needs to be adapted for the specific application domain. However, unaligned examples of text-images, and valid language sentences are abundant. With this in mind, we develop a method for text recognition which learns from such unaligned data. We cast the text recognition problem as one of aligning the conditional distribution of strings predicted from given text images, with lexically valid strings. This alignment is induced through an adversarial discriminator which tries to distinguish the predicted and real text strings apart. Our method achieves excellent text recognition accuracy, using no labelled training examples. Temporal sequences (videos) of objects encode changes in their pose. We develop a method to harness this, and learn object landmark detectors, which consistently track object parts across different poses and instances. We achieve this by conditionally generating a future frame given a past frame, and a sparse keypoint like (learnt) representation extracted from the future frame. We demonstrate generality of our method by learning landmarks for human faces (where we outperform existing landmark detectors), articulated human body, and rigid 3D objects, with no modification to the method. Finally, we propose one-step inductive training for improving generalisation in recurrent neural networks to longer sequences. We restrict the recurrent state to a spatial memory map which tracks the regions of the image which have been accounted for, and train the network for valid evolution of this map. We show excellent generalisation to much longer sequences on two sequential visual recognition tasks — joint localisation and recognition of multiple lines of text, and counting objects in aerial images.", "source_url": "https:\/\/openalex.org\/W2964327310", "pdf_url": "https:\/\/ora.ox.ac.uk\/objects\/uuid:0a9d2c4a-0e8d-4722-9e0b-3f247efa89e8\/files\/m9652124287993d30be22dafbd8178674", "doi": "10.5287\/ora-2nqvgy7gz", "handle": "", "repository_name": "Oxford University Research Archive (ORA) (University of Oxford)", "source_record_id": "W2964327310", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1042, "title": "Visual Aided Deep Learning Applications for Underwater Operations", "title_ar": "", "title_en": "", "author": "Martin Skaldebø", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Norway", "governorate": "", "publication_year": 2019, "language": "sv", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "deep-learning; computer-vision; generative-ai; robotics; ai-applications; cnn; object-detection; gan; ai-industry", "keywords": "", "abstract": "Den 4. industrielle revolusjonen har pågått de siste fire årene og resultater har begynt å vise seg i industrimarkedet. Google slipper stadig ny teknologi som kan slå mennesker i et mangfold av strategispill og de har utviklet kunstig inteligens som kan lære å gå på egenhånd. Statlige instanser og klassifikasjonsselskaper sammarbeider mot regler og retningslinjer for autonome skip, Yara Birkeland er et skip som er planlagt ferdig i 2020 som vil gå mot full autonom operasjon innen 2022. Autonomi i undervannsoperasjoner erfarer også vekst som vises igjen i økning i markedet. Maskinlæring er forventet å være en bidragsyter i skiftet mot en mer autonom fremtid. Metoder for innhenting av relevant informasjon i undervannsapplikasjoner som eksisterer idag bruker hovedsakelig akkustiske sensorer. Metoder for informasjonsinhenting som bruker kamerabilder er sjeldent brukt, men forbedringer innen nevrale nettverk, spesielt Convolutional Neural Networks (CNN), har vist lovende resultater. Denne rapporten vil undersøke bruken av slike metoder for undervannsoperasjoner. Rapporten trekker spesielt frem to problemstillinger, som er fordelt på to vedlagte artikler. Den første problemstillingen anngår realitetsgapet, som beskriver fenomenet som oppstår når kunnskap blir overført mellom domener, mer presist mellom det kunstige simuleringsdomenet og den virkelige verden. Eksisterende metoder angående dette har blitt undersøkt og sammenlignet for å danne et utsnitt av eksisterende metoder. Grunnet det tøffe miljøet under vann har bruken av kamerabilder til informasjonsinnhenting vært liten. Dårlig belysning, maritim snø, et miljø i kontinuerlig endring etc. gjør det vanskelig å modellere et undervannsmiljø. Det vil derfor nesten garantert eksistere et realitetsgap mellom det simulerte domenet og den virkelige verden. I løpet av de siste årene har flere rammeverk som angriper realitetsgapet oppstått og blitt publisert. Generative Adversarial Networks (GAN) har muligens stått frem som en av de mest lovende metodene for domenetransformasjon de siste årene. Denne rapporten undersøker en metode av dette rammeverket kalt CycleGAN, som introduserer en sykelkonsistens egenskap i det orginale GAN rammeverket. CycleGAN har vist gode resultater inne flere disipliner og har i denne rapporten for første gang blitt anvendt på undervannsapplikasjoner. Metoden består av å ta et bilde som input og reprodusere det samme bildet i et annet domene som output, e.g. ta et generert bilde fra simuleringsdata av et undervannspanel som input og produsere et virkelighetsnært bilde av det samme undervannspanelet. I denne rapporten er rammeverket anvendt på to forskjellige dataset med bilder fra undervannsmiljøer og en god kartlegging mellom simuleringsbaserte domener og den virkelige verden er oppnådd. Den andre problemstillingen omhandler optisk basert objekt detektering ved bruk av monokularkamera. Deep learning metoder for objekt detektering er undersøkt i unison med metoder for å ekstrakte relevante egenskaper fra de respektive detektorene for å bidra i et dynamsik posisjonerings (DP) system. En detekteringsalgoritme basert på YOLOv3 er anvendt på et dataset sammensatt av bilder fra marin kybernetisk laboratoriet (MC-lab) ved NTNU. Datasettet inkluderer bilder av et kjent objekt i bassenget i laboratoriet. Når detekteringsalgoritmen er ferdig trent og kan vellykket detektere det relevante objectet, er en skaleringsfunksjon samt et DP-system for undervannsdronen BlueROV utviklet. BlueROV2 utfører DP relativ til det kjente objektet ved å innhente relevant informasjon angående posisjon og avstander ved hjelp av detekteringsalgoritmen. Systemet fungerer tilfredstillende og klarer å utføre DP med korrekt informasjonsinnhenting av både posisjon og avstander til objektet fra kamerabilder.", "source_url": "https:\/\/openalex.org\/W2999495304", "pdf_url": "", "doi": "", "handle": "", "repository_name": "NORA - Norwegian Open Research Archives", "source_record_id": "W2999495304", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1045, "title": "Classification of ulcer images using convolutional neural networks", "title_ar": "", "title_en": "Classification of ulcer images using convolutional neural networks", "author": "Charlotta Aguirre Nilsson ; Medina Velic", "supervisor": "", "degree_type": "master", "degree_evidence": "نص السجل: «master»", "university": "Chalmers University of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; transfer-learning; cnn; image-classification; medical-imaging", "keywords": "", "abstract": "The use of artificial intelligence has increased within a variety of different fields the last decade, including the area of health care.Machine learning algorithms have already been successfully used in e.g.skin cancer detection in images, indicating its potential for being applied to wound treatment by automatically classifying wound images.In addition to reducing patient discomfort, a machine learning based solution for wound image classification could lower the costs for treatment of wounds.Today, this cost corresponds to the third biggest health care cost for Västra Götalandsregionen, the health care sector in Västra Götaland.This master thesis is a collaboration between the electronics and software development company QRTECH and Västra Götalandsregionen with the aim of creating an image-based aid to be used in the diagnosis of venous leg ulcers.A total of 300 manually annotated images depicting various ulcers for training a network have been provided by Alexandra Forssgren, specialist in dermatology and head of the dermatology department at Skaraborg Hospital.A deep learning based algorithm, using a pre-trained version of the VGG19 convolutional neural network, has been implemented to perform automated binary classification in order to separate venous leg ulcers from other ulcers.Since the annotated ulcer data set was limited, transfer learning was used by pre-training on ImageNet as well as on a set of dermoscopic skin images.The network was implemented in Python using the deep learning library Keras with TensorFlow as backend.To improve performance and reduce overfitting, methods such as dropout, L1\/L2 regularisation and data augmentation were applied.Augmentation included rotation, flipping, contrast enhancement as well as added noise and modified brightness.The best set-up was selected based on the performance metrics accuracy, precision and recall.For the test set, the final network reached an accuracy, precision and recall of approximately 85 %, 82 % and 75 %, respectively.Most certainly, the limited size of the data set affected the results negatively and a larger data set would increase the variety of the training examples, possibly leading to better performance.However, the performance metrics are still promising considering the small data set which proves the potential of a machine learning based solution for ulcer image classification.", "source_url": "https:\/\/openalex.org\/W2889535768", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Chalmers Publication Library (Chalmers University of Technology)", "source_record_id": "W2889535768", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1052, "title": "Mosquito Image Classification using Convolutional Neural Networks", "title_ar": "", "title_en": "Mosquito Image Classification using Convolutional Neural Networks", "author": "Sumanth Vissamsetty", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "medical-ai", "research_subdomain": "image-classification", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; ai-applications; cnn; image-classification; ai-cybersecurity", "keywords": "", "abstract": "Human life has always been affected by insects, especially mosquitoes, since it's early beginnings. This pesky insect acts as a vector that transmit pathogens through feeding on our blood, spreading life-threatening diseases like Zika Virus, Malaria, Dengue fever, Chikungunya and more. It is important to prevent these mosquitoes from harming humans and one way to do so is to control the mosquito population, or mosquito abatement as it is commonly known. It is important to note that not all mosquitoes are the same and each of them live, reproduce and attack in their own unique way. Hence it is crucial for humans to identify each of the mosquito species and study them in a detailed manner which turns out to be a very complicated and time-consuming problem that needs to be solved prior to any attempt at mosquito control. This gives rise to the need of novel algorithms to identify mosquitoes through image processing tasks, coupled with automated machine learning classification techniques. In our thesis, we have used Convolutional Neural Networks to build an image classification model to identify two life-stages of a mosquito, namely, Instar 4 and Pupae. Our algorithm yielded an average accuracy of over 95%, reducing the time to process similar classifications by hand to at least a hundredfold.", "source_url": "https:\/\/openalex.org\/W2965985756", "pdf_url": "https:\/\/repository.lsu.edu\/cgi\/viewcontent.cgi?article=5939&context=gradschool_theses", "doi": "10.31390\/gradschool_theses.4889", "handle": "", "repository_name": "Louisiana State University", "source_record_id": "W2965985756", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:57", "updated_at": "2026-11-15 16:52:57" }, { "id": 1066, "title": "Structured Deep Neural Networks for Speech Recognition", "title_ar": "", "title_en": "Structured Deep Neural Networks for Speech Recognition", "author": "Chunyang Wu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Cambridge", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "machine-learning; deep-learning; computer-vision; nlp; speech-recognition", "keywords": "", "abstract": "Deep neural networks (DNNs) and deep learning approaches yield state-of-the-art performance in a range of machine learning tasks, including automatic speech recognition. The multi-layer transformations and activation functions in DNNs, or related network variations, allow complex and difficult data to be well modelled. However, the highly distributed representations associated with these models make it hard to interpret the parameters. The whole neural network is commonly treated a ``black box''. The behaviours of activation functions and the meanings of network parameters are rarely controlled in the standard DNN training. Though a sensible performance can be achieved, the lack of interpretations to network structures and parameters causes better regularisation and adaptation on DNN models challenging. In regularisation, parameters have to be regularised universally and indiscriminately. For instance, the widely used L2 regularisation encourages all parameters to be zeros. In adaptation, it requires to re-estimate a large number of independent parameters. Adaptation schemes in this framework cannot be effectively performed when there are limited adaptation data. This thesis investigates structured deep neural networks. Special structures are explicitly designed, and they are imposed with desired interpretation to improve DNN regularisation and adaptation. For regularisation, parameters can be separately regularised based on their functions. For adaptation, parameters can be adapted in groups or partially adapted according to their roles in the network topology. Three forms of structured DNNs are proposed in this thesis. The contributions of these models are presented as follows. The first contribution of this thesis is the multi-basis adaptive neural network. This form of structured DNN introduces a set of parallel sub-networks with restricted connections. The design of restricted connectivity allows different aspects of data to be explicitly learned. Sub-network outputs are then combined, and this combination module is used as the speaker-dependent structure that can be robustly estimated for adaptation. The second contribution of this thesis is the stimulated deep neural network. This form of structured DNN relates and smooths activation functions in regions of the network. It aids the visualisation and interpretation of DNN models but also has the potential to reduce over-fitting. Novel adaptation schemes can be performed on it, taking advantages of the smooth property that the stimulated DNN offer. The third contribution of this thesis is the deep activation mixture model. Also, this form of structured DNN encourages the outputs of activation functions to achieve a smooth surface. The output of one hidden layer is explicitly modelled as the sum of a mixture model and a residual model. The mixture model forms an activation contour, and the residual model depicts fluctuations around this contour. The smoothness yielded by a mixture model helps to regularise the overall model and allows novel adaptation schemes.", "source_url": "https:\/\/openalex.org\/W2804696297", "pdf_url": "", "doi": "10.17863\/cam.23363", "handle": "", "repository_name": "Apollo (University of Cambridge)", "source_record_id": "W2804696297", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1071, "title": "Dysarthric Speech Recognition and Offline Handwriting Recognition using Deep Neural Networks", "title_ar": "", "title_en": "Dysarthric Speech Recognition and Offline Handwriting Recognition using Deep Neural Networks", "author": "Suhas Pillai", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Rochester Institute of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "deep-learning; computer-vision; nlp; medical-ai; cnn; rnn; image-classification; speech-recognition", "keywords": "", "abstract": "Millions of people around the world are diagnosed with neurological disorders like Parkinson’s, Cerebral Palsy or Amyotrophic Lateral Sclerosis. Due to the neurological damage as the disease progresses, the person suffering from the disease loses control of muscles, along with speech deterioration. Speech deterioration is due to neuro motor condition that limits manipulation of the articulators of the vocal tract, the condition collectively called as dysarthria. Even though dysarthric speech is grammatically and syntactically correct, it is difficult for humans to understand and for Automatic Speech Recognition (ASR) systems to decipher. With the emergence of deep learning, speech recognition systems have improved a lot compared to traditional speech recognition systems, which use sophisticated preprocessing techniques to extract speech features. In this digital era there are still many documents that are handwritten many of which need to be digitized. Offline handwriting recognition involves recognizing handwritten characters from images of handwritten text (i.e. scanned documents). This is an interesting task as it involves sequence learning with computer vision. The task is more difficult than Optical Character Recognition (OCR), because handwritten letters can be written in virtually infinite different styles. This thesis proposes exploiting deep learning techniques like Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) for offline handwriting recognition. For speech recognition, we compare traditional methods for speech recognition with recent deep learning methods. Also, we apply speaker adaptation methods both at feature level and at parameter level to improve recognition of dysarthric speech.", "source_url": "https:\/\/openalex.org\/W2612106303", "pdf_url": "https:\/\/scholarworks.rit.edu\/cgi\/viewcontent.cgi?article=10561&context=theses", "doi": "", "handle": "", "repository_name": "RIT Scholar Works (Rochester Institute of Technology)", "source_record_id": "W2612106303", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1080, "title": "Robust speech recognition in noisy and reverberant environments using deep neural network-based systems", "title_ar": "", "title_en": "Robust speech recognition in noisy and reverberant environments using deep neural network-based systems", "author": "Novoa Ilic ; José Eduardo", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Universidad de Chile", "university_source": "repository_host", "college": "", "department": "", "country": "Chile", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "machine-learning; deep-learning; nlp; robotics; speech-recognition", "keywords": "", "abstract": "In this thesis an uncertainty weighting scheme for deep neural network-hidden Markov model (DNN-HMM) based automatic speech recognition (ASR) is proposed to increase discriminability in the decoding process. To this end, the DNN pseudo-log-likelihoods are weighted according to the uncertainty variance assigned to the acoustic observation. The results presented here suggest that substantial reduction in word error rate (WER) is achieved with clean training. Moreover, modelling the uncertainty propagation through the DNN is not required and no approximations for non linear activation functions are made. The presented method can be applied to any network topology that delivers log likelihood-like scores. It can be combined with any noise removal technique and adds a minimal computational cost. This technique was exhaustively evaluated and combined with uncertainty-propagation-based schemes for computing the pseudo-log-likelihoods and uncertainty variance at the DNN output. Two proposed methods optimized the parameters of the weighting function by leveraging the grid search either on a development database representing the given task or on each utterance based on discrimination metrics. Experiments with Aurora-4 task showed that, with clean training, the proposed weighting scheme can reduce WER by a maximum of 21% compared with a baseline system with spectral subtraction and uncertainty propagation using the unscented transform. Additionally, it is proposed to replace the classical black box integration of automatic speech recognition technology in human-robot interaction (HRI) applications with the incorporation of the HRI environment representation and modeling, and the robot and user states and contexts. Accordingly, this thesis focuses on the environment representation and modeling by training a DNN-HMM based automatic speech recognition engine combining clean utterances with the acoustic channel responses and noise that were obtained from an HRI testbed built with a PR2 mobile manipulation robot. This method avoids recording a training database in all the possible acoustic environments given an HRI scenario. In the generated testbed, the resulting ASR engine provided a WER that is at least 26% and 38% lower than publicly available speech recognition application programming interfaces (APIs) with the loudspeaker and human speakers testing databases, respectively, with a limited amount of training data. This thesis demonstrates that even state-of-the-art DNN-HMM based speech recognizers can benefit by combining systems for which the acoustic models have been trained using different feature sets. In this context, the complementarity of DNN-HMM based ASR systems trained with the same data set but with different signal representations is discussed. DNN fusion methods based on flat-weight combination, the minimization of mutual information and the maximization of discrimination metrics were proposed and tested. Schemes that consider the combination of ASR systems with lattice combination and minimum Bayes risk decoding were also evaluated and combined with DNN fusion techniques. The experimental results were obtained using a publicly-available naturally-recorded highly reverberant speech data. Significant improvements in WER were observed by combining DNN-HMM based ASR systems with different feature sets, obtaining relative improvements of 10% with two classifiers and 18% with four classifiers, without any tuning or a priori information of the ASR accuracy.", "source_url": "https:\/\/openalex.org\/W2946330425", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Universidad de Chile", "source_record_id": "W2946330425", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1086, "title": "Single channel speech enhancement using Deep Neural Networks", "title_ar": "", "title_en": "Single channel speech enhancement using Deep Neural Networks", "author": "K S Kishor ; K. Sri Rama Murty", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Indian Institute of Technology Hyderabad", "university_source": "repository_host", "college": "", "department": "", "country": "India", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "machine-learning; deep-learning; nlp; speech-recognition", "keywords": "", "abstract": "Speech enhancement is an important first step in many applications like mobile communication, Speech recognition, hearing aids etc. Traditionally, speech enhancement was viewed as a pure signal processing problem, and several methods have been proposed to design linear filters to suppress the noise. However, with the advent of deep neural networks, speech enhancement has been viewed as a machine learning problem, which aims at learning a nonlinear model that maps the noisy speech signal to the clean speech signal. In this thesis, we provide an overview of existing signal processing approaches for single channel speech enhancement and compare their performance with the DNN counterparts. Even though the DNN based approaches provide significant performance improvements, they do not use phase information in the speech signal. In this work, We propose a speech enhancement framework using DNNs, which incorporates both magnitude (Cochleagram) and phase (Instantaneous frequency) information. Experimental results demonstrate that proposed framework achieves improved performance, especially at lower SNRs, over the conventional systems.", "source_url": "https:\/\/openalex.org\/W2764005420", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Research Archive of Indian Institute of Technology Hyderabad (Indian Institute of Technology Hyderabad)", "source_record_id": "W2764005420", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1087, "title": "Single-Channel Speech Enhancement Based on Deep Neural Networks", "title_ar": "", "title_en": "Single-Channel Speech Enhancement Based on Deep Neural Networks", "author": "Zhiheng Ouyang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Concordia University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "deep-learning; nlp; generative-ai; cnn; speech-recognition; gan", "keywords": "", "abstract": "Speech enhancement (SE) aims to improve the speech quality of the degraded speech. Recently, researchers have resorted to deep-learning as a primary tool for speech enhancement, which often features deterministic models adopting supervised training. Typically, a neural network is trained as a mapping function to convert some features of noisy speech to certain targets that can be used to reconstruct clean speech. These methods of speech enhancement using neural networks have been focused on the estimation of spectral magnitude of clean speech considering that estimating spectral phase with neural networks is difficult due to the wrapping effect. As an alternative, complex spectrum estimation implicitly resolves the phase estimation problem and has been proven to outperform spectral magnitude estimation. In the first contribution of this thesis, a fully convolutional neural network (FCN) is proposed for complex spectrogram estimation. Stacked frequency-dilated convolution is employed to obtain an exponential growth of the receptive field in frequency domain. The proposed network also features an efficient implementation that requires much fewer parameters as compared with conventional deep neural network (DNN) and convolutional neural network (CNN) while still yielding a comparable performance. Consider that speech enhancement is only useful in noisy conditions, yet conventional SE methods often do not adapt to different noisy conditions. In the second contribution, we proposed a model that provides an automatic \"on\/off\" switch for speech enhancement. It is capable of scaling its computational complexity under different signal-to-noise ratio (SNR) levels by detecting clean or near-clean speech which requires no processing. By adopting information maximizing generative adversarial network (InfoGAN) in a deterministic, supervised manner, we incorporate the functionality of SNR-indicator into the model that adds little additional cost to the system. We evaluate the proposed SE methods with two objectives: speech intelligibility and application to automatic speech recognition (ASR). Experimental results have shown that the CNN-based model is applicable for both objectives while the InfoGAN-based model is more useful in terms of speech intelligibility. The experiments also show that SE for ASR may be more challenging than improving the speech intelligibility, where a series of factors, including training dataset and neural network models, would impact the ASR performance.", "source_url": "https:\/\/openalex.org\/W3199713199", "pdf_url": "https:\/\/spectrum.library.concordia.ca\/986846\/1\/Ouyang_MASc_F2020.pdf", "doi": "", "handle": "", "repository_name": "Spectrum Research Repository (Concordia University)", "source_record_id": "W3199713199", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1088, "title": "Bayesian adaptation and combination of deep models for automatic speech recognition", "title_ar": "", "title_en": "Bayesian adaptation and combination of deep models for automatic speech recognition", "author": "Zhen Huang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Georgia Institute of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "deep-learning; nlp; generative-ai; speech-recognition", "keywords": "", "abstract": "The objective of the proposed research is to deploy a Bayesian adaptation and combination framework for deep model based automatic speech recognition systems to combat the degradation of the recognition accuracy, which is typically observed under potential mismatched conditions between training and testing. This dissertation addresses the problem in three directions. The first direction is to perform Bayesian adaptation directly on the discriminative deep neural network models. Maximum a posteriori estimation and multi-task learning techniques are employed in the manner of regularization in the deep neural network updating formula. In the second direction, deep neural network is cast into a generative model to better leverage Bayesian techniques. Classic structured maximum a posteriori adaption is adopted by using bottleneck features derived from deep neural networks. In the third direction, a hierarchical Bayesian system combination technique is employed to further enhance the adaptation performance by leveraging the complementarity of the discriminative and generative adaptive models.", "source_url": "https:\/\/openalex.org\/W2755331049", "pdf_url": "", "doi": "", "handle": "", "repository_name": "SMARTech Repository (Georgia Institute of Technology)", "source_record_id": "W2755331049", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1095, "title": "Transparency Detection in Computer Vision Applications", "title_ar": "", "title_en": "Transparency Detection in Computer Vision Applications", "author": "Volker Knauthe", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Technical University of Darmstadt", "university_source": "repository_host", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; computer-vision; expert-systems; ai-applications; object-detection; ai-industry", "keywords": "", "abstract": "Transparency is all around us. From windows to glass encased shopping malls. From drinking glasses to aquariums. And from safety goggles to protective encasings. It is one of the signs of technological advancement and prosperity, widely utilized in art and engineering. In contrast, it is not clearly understood how humans even perceive transparency. To build advanced autonomous moving agents or vision aids, we need to at least solve this intriguing challenge for computer vision applications. This thesis highlights and examines possible adverse effects that transparency related effects introduce to computer vision applications. Common depth estimation techniques only account for one depth per pixel, while a drinking glass already adds two possible depths. Looking through transparency, distortions and reflections affect object detection methods. Transparent objects transmit light differently, when dust accumulates or scratches interrupt their surface. This leads to gradually worse results from, e.g., condition monitoring setups behind protective glasses. Disregarding human made transparencies, there are also natural phenomena like heat shimmers and water droplets that play a role in industrial applications like condition monitoring or for autonomous agents. Solving those issues is not always possible, due to the ill posed task of restoring a pre transparency influenced image state. However, I introduce several possibilities of detecting and mitigating transparency related effects. A common issue with machine learning related solutions for novel challenges oftentimes lies in the scarcity of suitable data and\/or the massive labor cost involved in creating new data from scratch, as in the wider case of transparency. To enable further research concerning transparency related effects on computer vision applications, I introduce a state-of-the-art luminance keying dataset acquisition method that greatly facilitates the translation from real world objects into annotated datasets. It combines a black 99,99\\% light absorbing background with a post processing pipeline that leads to automated annotation masks. It is also the first easy setup that works with transparent objects, resolving the most demanding side effects from common chroma keying: no color bleeding, fuzzy edges, foreground background mixing, or shadows. In summary, this thesis showcases the importance and challenging aspects of transparency for computer vision applications and introduces novel prospects to facilitate further research.", "source_url": "https:\/\/openalex.org\/W7126018379", "pdf_url": "", "doi": "10.26083\/tuda-7751", "handle": "", "repository_name": "TUbilio (Technical University of Darmstadt)", "source_record_id": "W7126018379", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1096, "title": "Object detection and analysis using computer vision", "title_ar": "", "title_en": "Object detection and analysis using computer vision", "author": "Victor Friedmann Sandin ; Anna Thomsen", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "computer-vision; object-detection", "keywords": "", "abstract": "The project was performed at Consat Engineering AB’s facilities at Lindholmen, Gothenburg, in the spring of 2017. Consat Engineering AB produced a prototype that included a vision system, to analyze objects. This thesis exists to develop a proof of concept in form of a vision program, lowering the cost of the vision system, in order to be able to implement it in a commercial product. A pre-study comparing different hardware platforms and software libraries was conducted in order to best meet the requested specifications. The open source vision library OpenCV was chosen as the software library. A Raspberry Pi, running a Linux based OS called Raspbian, equipped with a camera was chosen as the hardware platform. Both of the choices included minimal expense towards the final cost of a future product that could be developed from this concept. After a finalized concept an evaluation test concluded that the program developed during this project fulfilled the requirements and proved that an open source solution significantly can lower the cost of the current vision system the prototype utilizes.", "source_url": "https:\/\/openalex.org\/W2772850197", "pdf_url": "", "doi": "", "handle": "", "repository_name": "", "source_record_id": "W2772850197", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1101, "title": "Automated Activity Discovery and Object Detection With Computer Vision: Towards Unsupervised Learning for Breakfast to Surgery", "title_ar": "", "title_en": "Automated Activity Discovery and Object Detection With Computer Vision: Towards Unsupervised Learning for Breakfast to Surgery", "author": "Michael Zhang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Harvard University", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; deep-learning; computer-vision; data-mining; predictive-analytics; cnn; object-detection; image-segmentation; video-analysis", "keywords": "", "abstract": "While deep learning in computer vision has enjoyed tremendous success in recent years, much progress remains to be made in important tasks. Many state-of-the-art vision models follow a supervised-learning paradigm, making them dependent on a large amount of labeled training data. This limits both their potential learning capabilities, as well as their application domains. In this work, we present contributions to expand the state-of-the-art in computer vision along both these dimensions, focusing on how to both perceive activities and objects in the video domain. We first tackle the problem of unsupervised activity discovery, where given a collection of untrimmed and unlabeled videos, we wish to learn a semantically meaningful embedding of the data which can be used to segment the data into \"discovered\" activities. To expand the learning capabilities of computer vision models, we avoid the intensive bottleneck of amassing a large annotated dataset, and seek to do so without requiring explicit training labels. We consider unsupervised activity discovery from the perspective of the inherently hierarchical nature of activities, e.g. a single complex activity may be modeled as a sequence of smaller sub-activities. Recognizing dependencies across this spectrum of complexities may therefore lead to greater video understanding. Accordingly, we introduce a hyperbolic embedding representation for video data to simultaneously capture hierarchical and semantic relationships in video data. While motivated by their prior success in modeling explicitly hierarchical data found in language, here we show how to leverage hyperbolic representations for the implicitly hierarchical nature of video data. We demonstrate that our hyperbolic video embeddings approach learn representations that significantly outperform the previous state-of-the-art for unsupervised activity segmentation on the Breakfast and 50Salads datasets, and that our hierarchical embeddings naturally allow discovery of activities at multiple levels of complexity. Following this, we next consider the second challenge of expanding the application domains of state-of-the-art computer vision models. We focus on open, or non-laparoscopic surgery, which represents the vast majority of all operating room procedures. Despite this prominence, few tools exist to objectively evaluate these techniques at scale, and current efforts involve human expert-based visual assessment. We therefore leverage a state-of-the-art convolutional neural network architecture for object detection to detect operating hands in open surgery videos. Automated assessment was expanded by combining model predictions with a fast object tracker to enable surgeon-specific hand tracking. To train our model, we used publicly available videos of open surgery from YouTube and annotated these with spatial bounding boxes of operating hands. Our model's spatial detections of operating hands significantly outperforms the detections achieved using pre-existing hand-detection datasets, and allow for insights into intra-operative movement patterns and economy of motion. Finally, we consider how to combine both advances in capability and application domain to focus on automatic activity discovery in surgery videos. Using our developed unsupervised learning algorithm, we demonstrate that we can discover multiple surgical activities in various operating procedures from just a collection of untrimmed and unlabeled YouTube surgery videos. We leverage a state-of-the-art deep learning architecture to extract base features, which we then encode into temporal and hierarchically-aware hyperbolic embeddings. We can then discover an arbitrary number of activities based on our clustering and decoding procedure. Through qualitative evaluation we show how our segmentations align with changes in video activity over time. Based on the alignment of our video segmentations, we present further insights and evidence into the hierarchical nature of activities. In all, our contributions present a step towards fully automated video activity understanding and discovery in real world domains.", "source_url": "https:\/\/openalex.org\/W3081564847", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Digital Access to Scholarship at Harvard (DASH) (Harvard University)", "source_record_id": "W3081564847", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1103, "title": "Towards Few-Annotation Learning in Computer Vision : Application to Image Classification and Object Detection tasks", "title_ar": "", "title_en": "", "author": "Quentin Bouniot", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2023, "language": "fr", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; transformers; image-classification; object-detection", "keywords": "", "abstract": "Vers un apprentissage avec peu d’annotations en vision par ordinateur : Application à la classification et à la détection d’objets dans les images Nous développons dans cette thèse des contributions théoriques, algorithmiques et expérimentales pour l’apprentissage automatique avec peu d'annotations, et plus spécifiquement pour les tâches de classification d’images et de détection d’objets en vision par ordinateur. Dans une première contribution, nous nous intéressons à combler le fossé entre la théorie et la pratique concernant les algorithmes de méta-apprentissage les plus répandus, utilisés pour de la classification avec peu d’exemples. Nous établissons des liens avec l’apprentissage de représentation multi-tâches, qui bénéficie de bases théoriques solides, afin de vérifier les meilleures conditions amenant à un méta-apprentissage plus efficace. Ensuite, afin d’exploiter les données non étiquetées pour entraîner des détecteurs d’objets basés sur l’architecture Transformer, nous proposons à la fois un pré-entraînement non supervisé et une méthode d’apprentissage semi-supervisée dans deux autres contributions distinctes. Pour le pré-entraînement, nous améliorons l’apprentissage contrastif des détecteurs d’objets en introduisant les informations de localisation. Enfin, notre méthode semi-supervisée est la première adaptée aux détecteurs d’objets utilisant des modèles type Transformers.", "source_url": "https:\/\/openalex.org\/W7151562911", "pdf_url": "", "doi": "10.70675\/6796455fzbdb0z4853z9023z932dbb24367b", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7151562911", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1105, "title": "Computer Vision for Distant Vehicle Detection: How to Find Region Proposals for Low-Resolution Objects?", "title_ar": "", "title_en": "Computer Vision for Distant Vehicle Detection: How to Find Region Proposals for Low-Resolution Objects?", "author": "Ann-Katrin Fattal", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "computer-vision", "research_subdomain": "cnn", "categories": "machine-learning; deep-learning; computer-vision; cnn", "keywords": "", "abstract": "Safety is crucial to the development and acceptance of assisted and highly automated driving functions. In 2017, 69.3% of German fatal accidents happened on roads where the speed limit was not enforced or higher than 100km\/h. At this speed, to perform safe driving maneuvers, the environment perception is a key element. Detecting objects in distances up to 200m is instrumental in anticipating potential obstacles. Due to hardware limitations, an automotive camera maps cars in e.g. 200m distance to an image of only 8px width. Hence, the absence of local details degrades the state-of-the-art detection methods designed for detecting bigger sized objects. The scope of this thesis is to develop, extend and evaluate object region localizers to improve the detection range of cameras. A saliency inspired voting map is proposed that highlights anomalies in automotive scenes. The environment is modeled with few homogeneous regions representing the background within the image. Such global features allow detecting small object regions. Inspired by the concept of learning features, this thesis presents machine learning methods detecting small objects. Existing labeled data sets such as the KITTI data set only have object regions which sizes are larger than 25px height. The presented methods in this thesis are performed against a newly created data set with 67% of object regions having a width of 8-30px, a range that has rarely been subject to research yet. Convolutional Neural Network based localizers have been evaluated and extended. To maintain a low computational power, only small networks can be used. However, such networks are limited to the usage of local features. An incorporation of global generic priors to local networks is proposed, which increases the recall especially for small object regions. The parameters to adjust Region Proposal Networks (RPNs) for the special case of small objects are further optimized and the main parameters are identified. A novel relevance based net-surgery is introduced, allowing to select the most relevant features while maintaining the recall of the RPN. It is then possible to reduce the network size to these few features.", "source_url": "https:\/\/openalex.org\/W3014231450", "pdf_url": "https:\/\/tuprints.ulb.tu-darmstadt.de\/11310\/13\/2019-10-30_Fattal_AnnKatrin.pdf", "doi": "10.25534\/tuprints-00011310", "handle": "", "repository_name": "Technischen Universität Darmstadt", "source_record_id": "W3014231450", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1110, "title": "Study for the vision-based object detection and tracking on unmanned aerial vehicles", "title_ar": "", "title_en": "Study for the vision-based object detection and tracking on unmanned aerial vehicles", "author": "Eduard Rius Martínez", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Universitat Politècnica de Catalunya", "university_source": "repository_host", "college": "", "department": "", "country": "Spain", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "computer-vision; robotics; object-detection", "keywords": "", "abstract": "This document is a degree thesis that consists of a study of vision-based computing techniques for the detection and tracking of objects. The development of the study has been carried out using OpenCV, CUDA and PX4. The first chapter presents a definition of the objectives, requirements and planning for the development of the study. Next a brief introduction to computer vision and its state of the art is included. After presenting the work to be done, chapters 3, 4 and 5 explain the different approaches to detection, tracking and control algorithms, and each chapter explains the solution chosen for each one. Chapter 6 introduces the whole system and explains how the above algorithms are implemented in a drone. Finally, in chapter 7 the tests and their results are presented and discussed", "source_url": "https:\/\/openalex.org\/W3165911597", "pdf_url": "", "doi": "", "handle": "", "repository_name": "UPCommons institutional repository (Universitat Politècnica de Catalunya)", "source_record_id": "W3165911597", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1113, "title": "STABILIZATION AND OBJECT TRACKING BASED ON COMPUTER VISION: APPLICATION IN A QUAD-ROTOR", "title_ar": "", "title_en": "STABILIZATION AND OBJECT TRACKING BASED ON COMPUTER VISION: APPLICATION IN A QUAD-ROTOR", "author": "Luis Arreola", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; computer-vision; robotics; object-detection", "keywords": "", "abstract": "\"This work presents a Research and Development thesis of a functional Unmanned Aerial Vehicle (UAV), integrating optical flow algorithms and navigation based on object detection using Computer-Vision Features Classifiers and Robotics. This work presents two main problems to solve. In one of them, a position estimation system for Unmanned Aerial Vehicles consisting of hardware and software is presented. It is based on low-cost devices(GPS, commercial autopilot sensors and optical flow algorithm) implemented in an on-board microcomputer. Comparative tests were conducted using the proposed approach and the conventional one, where only fusion of GPS and inertial sensors are used. Furthermore, the second problem addresses object detection and tracking based on Computer Vision techniques, in essence, a training technique known as Haar Feature-based cascade classifier, image processing is made on-board with a high processor single-board computer. Based on the detected object and its position, the UAV must track it to ensure to be positioned in a center position with respect to the object and in a safe distance to it.\"", "source_url": "https:\/\/openalex.org\/W2902442432", "pdf_url": "", "doi": "", "handle": "", "repository_name": "LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas)", "source_record_id": "W2902442432", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1114, "title": "Discourse-aware neural machine translation", "title_ar": "", "title_en": "Discourse-aware neural machine translation", "author": "Wang Longyue", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Dublin City University", "university_source": "repository_host", "college": "", "department": "", "country": "Ireland", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "nlp", "research_subdomain": "machine-translation", "categories": "machine-learning; nlp; predictive-analytics; machine-translation", "keywords": "", "abstract": "Machine translation (MT) models usually translate a text by considering isolated sentences based on a strict assumption that the sentences in a text are independent of one another. However, it is a truism that texts have properties of connectedness that go beyond those of their individual sentences. Disregarding dependencies across sentences will harm translation quality especially in terms of coherence, cohesion, and consistency. Previously, some discourse-aware approaches have been investigated for conventional statistical machine translation (SMT). However, this is a serious obstacle for the state-of-the-art neural machine translation (NMT), which recently has surpassed the performance of SMT. In this thesis, we try to incorporate useful discourse information for enhancing NMT models. More specifically, we conduct research on two main parts: 1) exploiting novel document-level NMT architecture; and 2) dealing with a specific discourse phenomenon for translation models. Firstly, we investigate the influence of historical contextual information on the perfor- mance of NMT models. A cross-sentence context-aware NMT model is proposed to consider the influence of previous sentences in the same document. Specifically, this history is summarized using an additional hierarchical encoder. The historical representations are then integrated into the standard NMT model in different strategies. Experimental results on a Chinese–English document-level translation task show that the approach significantly improves upon a strong attention-based NMT system by up to +2.1 BLEU points. In addition, analysis and comparison also give insightful discussions and conclusions for this research direction. Secondly, we explore the impact of discourse phenomena on the performance of MT. In this thesis, we focus on the phenomenon of pronoun-dropping (pro-drop), where, in pro-drop languages, pronouns can be omitted when it is possible to infer the referent from the context. As the data for training a dropped pronoun (DP) generator is scarce, we propose to automatically annotate DPs using alignment information from a large parallel corpus. We then introduce a hybrid approach: building a neural-based DP generator and integrating it into the SMT model. Experimental results on both Chinese–English and Japanese–English translation tasks demonstrate that our approach achieves a significant improvement of up to +1.58 BLEU points with 66% F-score for DP generation accuracy. Motivated by this promising result, we further exploit the DP translation approach for advanced NMT models. A novel reconstruction-based model is proposed to reconstruct the DP-annotated source sentence from the hidden states of either encoder or decoder, or both components. Experimental results on the same translation tasks show that the proposed approach significantly and consistently improves translation performance over a strong NMT baseline, which is trained on DP-annotated parallel data. To avoid the errors propagated from an external DP prediction model, we finally investigate an end-to-end DP translation model. Specifically, we improve the reconstruction-based model from three perspectives. We first employ a shared reconstructor to better exploit encoder and decoder representations. Secondly, we propose to jointly learn to translate and predict DPs. In order to capture discourse information for DP prediction, we finally combine the hierarchical encoder with the DP translation model. Experimental results on the same translation tasks show that our approach significantly improves both translation performance and DP prediction accuracy.", "source_url": "https:\/\/openalex.org\/W2951604644", "pdf_url": "https:\/\/doras.dcu.ie\/22903\/1\/Longyue_PhD_Thesis_Final.pdf", "doi": "", "handle": "", "repository_name": "Dublin City University Open Access Institutional Repository (Dublin City University)", "source_record_id": "W2951604644", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1116, "title": "Understanding, Explaining, and Aligning Neural Models for Machine Translation", "title_ar": "", "title_en": "Understanding, Explaining, and Aligning Neural Models for Machine Translation", "author": "Patrick Fernandes", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; nlp; predictive-analytics; llm; machine-translation", "keywords": "", "abstract": "The recent success of large neural network models trained on massive amounts of unlabeled data has, yet again, highlighted the power of scaling, in terms of both model capacity and data quantity. However, despite its successes, scaling up also adds new additional challenges: (i) the large size and complexity of these models make it difficult to understand their learning dynamics and global behavior, and how these change with the data distribution; (ii) their complexity also make it difficult to extract explanations of how they arrive at individual predictions, and to know how to leverage these explanations; (iii) this, coupled with the fact that the data used to train these models is often noisy, can lead to models that do not align with their intended use case. This thesis tackles all of the challenges above, with a particular focus on models trained for machine translation (MT). In the first part, we study the scaling behaviour of MT mod els trained to translate between multiple high-resource language pairs (LPs), showing that the way performance for individual LPs scales with model capacity is only dependent on the properties of that “task” and is invariant to cross-lingual interactions. We also investigate howmuchcontextinformationMTmodelsleverageindocument-levelMT,proposingways to increase it. In the second part, we start by proposing a novel method for extracting expla nations of how neural models arrive at their predictions, relying on the assumption that good explanations should help other models learn, and leveraging bilevel optimization\/meta learning to learn explainers that teach well. Then we showthat not onlyare reasoning models state-of-the-art systems for low-resource MT, but that their chain-of-thought rationales are good explanations to help teach smaller LLMs the same task. In the third and final part, we start by showing that, by including metrics of translation quality (trained on human feedback annotations) in the inference stage of a trained MTmodel, we can improve the qualityof the translation and align the system towards human-like translations. We then also show that we can almost completely remove the dependency on human annotations by leveraging large language models to provide (fine-grained) translation quality feedback.", "source_url": "https:\/\/openalex.org\/W7116361097", "pdf_url": "", "doi": "10.1184\/r1\/30752876", "handle": "", "repository_name": "KiltHub Repository", "source_record_id": "W7116361097", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1119, "title": "Neural Language Models with Morphology for Machine Translation", "title_ar": "", "title_en": "Neural Language Models with Morphology for Machine Translation", "author": "Tomáš Musil", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Czech Republic", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "deep-learning; nlp; llm; machine-translation", "keywords": "", "abstract": "Language models play an important role in many natural language processing tasks. In this thesis, we focus on language models built on artificial neural net- works. We examine the possibilities of using morphological annotations in these models. We propose a neural network architecture for a language model that explicitly makes use of morphological annotation of the input sentence: instead of word forms it processes lemmata and morphological tags. Both the baseline and the proposed method are evaluated on their own by perplexity, and also in the context of machine translation by the means of automatic translation quality evaluation. While in isolation the proposed model significantly outperforms the baseline, there is no apparent gain in machine translation. 1", "source_url": "https:\/\/openalex.org\/W2947679756", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Digital Repository (National Repository of Grey Literature)", "source_record_id": "W2947679756", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1122, "title": "Domain adaptation for statistical machine translation and neural machine translation", "title_ar": "", "title_en": "Domain adaptation for statistical machine translation and neural machine translation", "author": "Jian Zhang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Dublin City University", "university_source": "repository_host", "college": "", "department": "", "country": "Ireland", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; nlp; transfer-learning; llm; machine-translation", "keywords": "", "abstract": "Both Statistical Machine Translation and Neural Machine Translation (NMT) are data-dependent learning approaches to Machine Translation (MT). The prerequisite is a large volume of training data in order to generate good statistical models. However, even if large volume of training corpora are available for MT, finding training data which are similar to the specific domains is still difficult. The MT models trained using the limited specific domain data cannot have sufficient coverage on the linguistic phenomena in that domain, which makes this a very challenging task. Because word meanings, genres or topics differ between domains, using the additional data from other domains can increase the dissimilar- ities between the training and testing data, and result in reduced translation quality. Such a challenge is defined as ‘domain adaptation’ challenge in the literature. In this thesis, we investigate domain adaptation in two different scenarios, namely a domain-awareness scenario and a domain-unawareness scenario. In a domain-awareness scenario, the domain information is given explicitly in the training data. We are interested in developing domain-adaptation techniques which transfer knowledge gained from the other domains to a desired domain. In the approach proposed here probabilistic values indicating the domain-likeness features for words are estimated by the context rather than by the words themselves. We then apply those features to the combined translation models in an MT system. We empirically show that translation quality can be significantly improved compared with previous related work. We then turn our interest to the recently proposed neural network training. We describe a domain-adaptation approach which can exploit large pre-trained word vector models. We evaluate our approach on both language modelling and machine translation tasks to demonstrate its efficiency, effectiveness and flexibility in a domain-awareness scenario. xiiIn a domain-unawareness scenario, the domain information is not given explicitly in the training data. The training data is heterogeneous, e.g. originated from tens or even hundreds of different resources without well-defined domain labels. We overcome such a challenge by deriving the topic information from the training corpora using well-estimated topic modelling algorithms. In this scenario, we pay particular attention to the most recent NMT framework. We are concerning with making a better lexical choice and improving the overall translation quality. Experimentally, we show that our model can perform better lexical choice, improve the overall translation quality and reduce the number of unknown words.", "source_url": "https:\/\/openalex.org\/W2767972135", "pdf_url": "https:\/\/doras.dcu.ie\/21949\/1\/thesis_hardbound_jian_zhang_56101775.pdf", "doi": "", "handle": "", "repository_name": "Dublin City University Open Access Institutional Repository (Dublin City University)", "source_record_id": "W2767972135", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1134, "title": "Neural machine translation with limited resources", "title_ar": "", "title_en": "Neural machine translation with limited resources", "author": "Tasnim Mohiuddin", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Nanyang Technological University", "university_source": "repository_host", "college": "", "department": "", "country": "Singapore", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "nlp", "research_subdomain": "machine-translation", "categories": "machine-learning; deep-learning; computer-vision; nlp; transfer-learning; predictive-analytics; machine-translation", "keywords": "", "abstract": "With the advent of deep neural networks in recent years, Neural Machine Translation (NMT) systems have achieved state-of-the-art performance on standard translation benchmarks. NMT is a way to translate from one language to another with a single neural network in an end-to-end manner. The NMT models have emerged quickly, and within a few years of research, they have outperformed the traditional statistical systems with impressive performance. Despite the success of NMT models in standard benchmarks, there are some notable limitations. One of them is that NMT models are known to be data-hungry, i.e., they tend to work very well only when a massive amount of parallel training data (a.k.a. bitext) is available, but perform poorly when the data is limited. Except for some mainstream languages, e.g., English, French, or Chinese, most natural languages are low-resourced and lack large parallel data. Moreover, acquiring large bitext corpora is not viable in most scenarios, especially with resource-constrained conditions like low-resource languages. Researchers have made numerous endeavors to expand the success of NMT from high-resource to low-resource languages like transfer learning, data augmentation, and pivoting. However, they still require strong cross-lingual signals, i.e., lots of parallel data. One solution to this problem might be transferring cross-lingual signals through cross-lingual word embeddings (CLWEs), which can be learned from monolingual data in an unsupervised way or with the help of a small seed dictionary. CLWEs seem to be very promising in resource-constrained machine translation (MT). Most of the successful and predominant CLWE methods (a.k.a. word translation methods) learn a linear mapping function based on the isomorphic assumption, which is problematic. We hypothesize to learn the cross-lingual mapping in a projected latent space which would give the model enough flexibility to induce the required geometric structures such that it would be easier to align the embeddings. Based on this hypothesis, we propose two novel models for learning CLWEs. We empirically show that our methods are particularly very effective for low-resource languages. We then turn our attention from word- to sentence-level translation with limited resources. Specifically, we focus on data augmentation strategies widely used in NLP and Computer Vision to increase the robustness of the models in resource-constrained scenarios. We investigate the domain-mismatch issue thoroughly that hinders the all-embracing success of the existing techniques in NMT. Eventually, we introduce a novel data augmentation framework for low-resource NMT that leverages the neighboring samples of the original parallel data without explicitly using additional monolingual data. Our framework can diversify the in-domain parallel data in a controlled way. We perform extensive experiments on four low-resource language pairs comprising data from different domains. We have shown that our method is comparable to the traditional back-translation that uses extra in-domain monolingual data. Typically, NMT systems are trained on heterogeneous data from different domains, sources, topics, styles, and modalities. The quality of the data also varies a lot. Usually, during training, all the data are concatenated and randomly shuffled. However, not all of them may be useful, some data may be redundant, and some might even be noisy and detrimental to the final NMT system performance. These problems are more acute in low-resource languages compared to the high-resource ones. Consequently, we explore the possibilities of curriculum training for NMT systems, i.e., presenting the data to the NMT systems in a systematic order during training. We introduce a two-stage curriculum training framework for NMT where we fine-tune a base NMT model on subsets of data. To select the data subsets, we propose two scoring approaches --- deterministic scoring using pre-trained methods and online scoring that considers prediction scores of the emerging NMT model. Our curriculum strategies consistently demonstrate better translation quality and faster convergence (approximately 50% fewer updates) on both high- and low-resource languages.", "source_url": "https:\/\/openalex.org\/W4281623321", "pdf_url": "https:\/\/dr.ntu.edu.sg\/bitstream\/10356\/157475\/2\/Thesis-Tasnim.pdf", "doi": "10.32657\/10356\/157475", "handle": "", "repository_name": "DR-NTU (Nanyang Technological University)", "source_record_id": "W4281623321", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1136, "title": "Non-Autoregressive Neural Machine Translation", "title_ar": "", "title_en": "Non-Autoregressive Neural Machine Translation", "author": "Jindřich Helcl", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Czech Republic", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "nlp", "research_subdomain": "machine-translation", "categories": "deep-learning; nlp; machine-translation", "keywords": "", "abstract": "In recent years, a number of mehtods for improving the decoding speed of neural machine translation systems have emerged. One of the approaches that pro- poses fundamental changes to the model architecture are non-autoregressive models. In standard autoregressive models, the output token distributions are conditioned on the previously decoded outputs. The conditional dependence al- lows the model to keep track of the state of the decoding process, which improves the fluency of the output. On the other hand, it requires the neural network computation to be run sequentially, and thus it cannot be parallelized. Non- autoregressive models impose conditional independence on the output distri- butions, which means that the decoding process is parallelizable and hence the decoding speed improves. A major drawback of this approach is lower trans- lation quality compared to the autoregressive models. The goal of the non- autoregressive translation research is to find methods that improve the trans- lation quality, while retaining high decoding speed. In this thesis, we explore the research progress so far and identify flaws in the generally accepted eval- uation methodology. We experiement with non-autoregressive models trained with connectionist temporal classification. We find that even though our models...", "source_url": "https:\/\/openalex.org\/W7135983976", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Digital Repository (National Repository of Grey Literature)", "source_record_id": "W7135983976", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1139, "title": "Intelligent data mining using artificial neural networks and genetic algorithms : techniques and applications", "title_ar": "", "title_en": "Intelligent data mining using artificial neural networks and genetic algorithms : techniques and applications", "author": "Jianhua Yang", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Warwick", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "xai", "categories": "machine-learning; deep-learning; medical-ai; responsible-ai; expert-systems; data-mining; predictive-analytics; optimization; xai", "keywords": "", "abstract": "Data Mining (DM) refers to the analysis of observational datasets to find relationships and to summarize the data in ways that are both understandable and useful. Many DM techniques exist. Compared with other DM techniques, Intelligent Systems (ISs) based approaches, which include Artificial Neural Networks (ANNs), fuzzy set theory, approximate reasoning, and derivative-free optimization methods such as Genetic Algorithms (GAs), are tolerant of imprecision, uncertainty, partial truth, and approximation. They provide flexible information processing capability for handling real-life situations. This thesis is concerned with the ideas behind design, implementation, testing and application of a novel ISs based DM technique. The unique contribution of this thesis is in the implementation of a hybrid IS DM technique (Genetic Neural Mathematical Method, GNMM) for solving novel practical problems, the detailed description of this technique, and the illustrations of several applications solved by this novel technique. GNMM consists of three steps: (1) GA-based input variable selection, (2) Multi- Layer Perceptron (MLP) modelling, and (3) mathematical programming based rule extraction. In the first step, GAs are used to evolve an optimal set of MLP inputs. An adaptive method based on the average fitness of successive generations is used to adjust the mutation rate, and hence the exploration\/exploitation balance. In addition, GNMM uses the elite group and appearance percentage to minimize the randomness associated with GAs. In the second step, MLP modelling serves as the core DM engine in performing classification\/prediction tasks. An Independent Component Analysis (ICA) based weight initialization algorithm is used to determine optimal weights before the commencement of training algorithms. The Levenberg-Marquardt (LM) algorithm is used to achieve a second-order speedup compared to conventional Back-Propagation (BP) training. In the third step, mathematical programming based rule extraction is not only used to identify the premises of multivariate polynomial rules, but also to explore features from the extracted rules based on data samples associated with each rule. Therefore, the methodology can provide regression rules and features not only in the polyhedrons with data instances, but also in the polyhedrons without data instances. A total of six datasets from environmental and medical disciplines were used as case study applications. These datasets involve the prediction of longitudinal dispersion coefficient, classification of electrocorticography (ECoG)\/Electroencephalogram (EEG) data, eye bacteria Multisensor Data Fusion (MDF), and diabetes classification (denoted by Data I through to Data VI). GNMM was applied to all these six datasets to explore its effectiveness, but the emphasis is different for different datasets. For example, the emphasis of Data I and II was to give a detailed illustration of how GNMM works; Data III and IV aimed to show how to deal with difficult classification problems; the aim of Data V was to illustrate the averaging effect of GNMM; and finally Data VI was concerned with the GA parameter selection and benchmarking GNMM with other IS DM techniques such as Adaptive Neuro-Fuzzy Inference System (ANFIS), Evolving Fuzzy Neural Network (EFuNN), Fuzzy ARTMAP, and Cartesian Genetic Programming (CGP). In addition, datasets obtained from published works (i.e. Data II & III) or public domains (i.e. Data VI) where previous results were present in the literature were also used to benchmark GNMM’s effectiveness. As a closely integrated system GNMM has the merit that it needs little human interaction. With some predefined parameters, such as GA’s crossover probability and the shape of ANNs’ activation functions, GNMM is able to process raw data until some human-interpretable rules being extracted. This is an important feature in terms of practice as quite often users of a DM system have little or no need to fully understand the internal components of such a system. Through case study applications, it has been shown that the GA-based variable selection stage is capable of: filtering out irrelevant and noisy variables, improving the accuracy of the model; making the ANN structure less complex and easier to understand; and reducing the computational complexity and memory requirements. Furthermore, rule extraction ensures that the MLP training results are easily understandable and transferrable.", "source_url": "https:\/\/openalex.org\/W2111157768", "pdf_url": "http:\/\/wrap.warwick.ac.uk\/3831\/1\/WRAP_THESIS_Yang_2010.pdf", "doi": "10.82444\/warw.33491953", "handle": "", "repository_name": "Warwick Research Archive Portal (University of Warwick)", "source_record_id": "W2111157768", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1142, "title": "Prediction system for Cardio-vascular diseases using data mining algorithms.", "title_ar": "", "title_en": "Prediction system for Cardio-vascular diseases using data mining algorithms.", "author": "Tinotenda Mavetera", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "North-west University", "university_source": "repository_host", "college": "", "department": "", "country": "South Africa", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "medical-ai", "research_subdomain": "data-mining", "categories": "machine-learning; deep-learning; medical-ai; data-mining; predictive-analytics", "keywords": "", "abstract": "Background: Cardiovascular disease (CVD) is a major global health concern, accounting for a significant proportion of morbidity and mortality worldwide. Early detection and prevention of CVD reduces the burden of this disease, and data mining techniques show great potential in predicting CVD risk. Data mining algorithms use large datasets to identify patterns and relationships that may not be immediately apparent, making them specifically suitable for analysing complex biomedical data. Objective: This research is designed to determine the best performing machine learning algorithm based on the dataset and demonstrate how this algorithm could be used to predict cardiovascular disease and improve both diagnosis and treatment. The Study examined various data mining techniques and their performance in predicting CVD risk and explore the potential of combining different algorithms for improved prediction accuracy. Methodology: In this research, Knowledge Discovery in Database with iterative and interactive techniques was used to identify patterns from the dataset. The dataset used in this research was obtained from UCI Machine Learning Repository. Results: The results of this study demonstrate that three out of the four implemented algorithms exhibited excellent performance, yielding high classification accuracy for predicting cardiovascular disease. Specifically, Naïve Bayes achieved 85.25%, Decision Tree achieved 71.43%, while the Artificial Neural Network outperformed the other algorithms with a classification accuracy of 91.75%. A classification accuracy of 67% achieved by KNN cannot be said to be excellent. These findings suggest that the top three algorithms could be utilized effectively for predicting the presence of cardiovascular disease. Conclusion: To conclude, this research demonstrates that machine learning could be applied to healthcare and assist medical practitioners to diagnose the presence of cardiovascular disease effectively and thus contribute to the development of efficient CVD prevention strategies that improve public health outcomes.", "source_url": "https:\/\/openalex.org\/W7225216488", "pdf_url": "https:\/\/repository.nwu.ac.za\/bitstreams\/d117cde3-cd54-4fd4-b347-469073b0c890\/download", "doi": "", "handle": "", "repository_name": "Boloka Institutional Repository (North-west University)", "source_record_id": "W7225216488", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1152, "title": "PENERAPAN DATA MINING UNTUK MEMPREDIKSI PENJUALAN HANDPHONE PADA KONTER A CELL MENGGUNAKAN ALGORITMA BAYESIAN CLASSIFICATION", "title_ar": "", "title_en": "PENERAPAN DATA MINING UNTUK MEMPREDIKSI PENJUALAN HANDPHONE PADA KONTER A CELL MENGGUNAKAN ALGORITMA BAYESIAN CLASSIFICATION", "author": "Hevin Saputra", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "machine-learning", "research_subdomain": "data-mining", "categories": "machine-learning; data-mining; predictive-analytics", "keywords": "data mining, algorithm of bayesian classification, the number of sales of mobile", "abstract": "The background of this research is the number of sales of electronic phone products in A Cell in Sleman Yogyakarta which is often uncertain. The company has a lot of sales transaction data stored in Ms. Excel. The data only serves as an archive, so the sale on A Cell decreases because it cannot predict the data to be used as A decision maker. The application of Data Mining method is very necessary to calculate the classification of the number of mobile phone sales that function get the percentage result of accuracy of each brand of handphone. The system can improve and build systems that are still less than optimal. In evaluating the number of mobile phone sales implemented the implementation of data mining using Bayesian classification algorithm to get the best transaction data from the results of the sales of mobile phone products that have been in prediction. Therefore, through the application of data mining using Bayesian classification algorithm is expected by the method can be used as a sales strategy to assist in increasing mobile phone sales. Keywords: data mining, algorithm of bayesian classification, the number of sales of mobile", "source_url": "https:\/\/openalex.org\/W2991485768", "pdf_url": "", "doi": "", "handle": "", "repository_name": "", "source_record_id": "W2991485768", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:58", "updated_at": "2026-11-15 16:52:58" }, { "id": 1165, "title": "A Framework for Explainable Artificial Intelligence in Geospatial Contexts", "title_ar": "", "title_en": "A Framework for Explainable Artificial Intelligence in Geospatial Contexts", "author": "Cédric Roussel", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Germany", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "computer-vision", "research_subdomain": "xai", "categories": "machine-learning; computer-vision; responsible-ai; ai-applications; predictive-analytics; xai; ai-smart-cities", "keywords": "", "abstract": "Artificial intelligence has become a key driver of innovation in geospatial data analysis. It enables new prediction, classification and decision-making methods in areas such as urban planning, environmental monitoring and mobility services. However, many artificial intelligence models, particularly complex machine learning algorithms, lack transparency. This makes it difficult for users to understand, validate or trust the outcomes of these models. This is particularly problematic in geospatial contexts, where decisions often have real-world implications and depend on local knowledge and geospatial reasoning. While existing explainable artificial intelligence methods offer insights into model behavior, they often disregard the geospatial context, such as the location of phenomena and the relationships between nearby areas. This limits their effectiveness in understanding and interpreting the geospatial behavior of the model. This dissertation addresses this challenge by advancing the field of geospatial explainable artificial intelligence, an emerging area that redefines how explanations relating to geospatial data should be constructed and communicated. It proposes a novel framework that works with a new type of explanation which can be tailored to the geospatial context. Glocal explanations are computed through the aggregation of local explanations and defined conditions. This enables explanations to be created that range between local and global explainable artificial intelligence. As for explanations, the model quality is adapted to these glocal units, enabling analysis of where and when the model performs well or poorly. The framework also provides interactive visualizations that enable users to explore, interpret, and question model outputs. One visualization is an interactive matrix, while a second uses multi-level glyphs to enable geospatial analysis of the explanations. Both visualizations integrate model quality to reliably interpret the reasoning behind the model's decisions. The novel contributions are supported by real-world experiments and prototype implementations. Three distinct use cases are considered: predicting rental bike station bookings, predicting car park occupancy, and classifying accident severity. An additional evaluation of the usability of the visualizations shows that representations such as the developed matrix or geovisualization significantly improve efficiency, effectiveness, and satisfaction. The final framework can be used by different types of users. Machine learning experts can use it to identify and understand where and why the model performs poorly. Domain experts can use the framework to gain a better understanding of the actions and causes within a use case, potentially enabling them to take actions that could have economic and social benefits.", "source_url": "https:\/\/openalex.org\/W7163335674", "pdf_url": "", "doi": "10.13140\/rg.2.2.17271.79525", "handle": "", "repository_name": "Fraunhofer-Publica (Fraunhofer-Gesellschaft)", "source_record_id": "W7163335674", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1168, "title": "Reliability Assessment Methods for Explainable Artificial Intelligence", "title_ar": "", "title_en": "", "author": "Sara Narteni", "supervisor": "", "degree_type": "phd", "degree_evidence": "نص السجل: «PhD»", "university": "", "university_source": "", "college": "", "department": "", "country": "Portugal", "governorate": "", "publication_year": 2025, "language": "it", "research_domain": "computer-vision", "research_subdomain": "xai", "categories": "computer-vision; responsible-ai; expert-systems; ai-applications; xai; ai-education; ai-cybersecurity", "keywords": "", "abstract": "Nel ripensare alla me stessa di ormai quasi 5 anni fa, appena laureata ed un po' \"smarrita\" rispetto alle diverse strade che potessi avere di fronte, mi piace sempre ricordare di come, quasi per caso, mi fossi affacciata al mondo della ricerca, intraprendendo questo percorso di dottorato.Compiere questa scelta ha significato per me non solo affrontare tematiche scientifiche quasi completamente nuove, imparando nuovi argomenti e soprattutto imparando ad apprenderli, ma mettermi alla prova con quanto di più avessi sempre ritenuto essere troppo al di là della mia zona di comfort: dal presentare le mie ricerche in pubblico, al partecipare ad eventi per \"semplice\" networking.Ed è così che questi 3 anni mi hanno permesso di crescere sotto il punto di vista professionale, trovando una direzione che la me neolaureata non sentiva di avere, ma anche dal punto di vista personale.Grazie al dottorato, posso dire -sono sincera, non senza una buona dose di orgoglio -di avere fatto un importante passo in avanti nel migliorare quelli che sono sempre stati un po' i miei limiti personali: avere parlato in pubblico ormai non so quante volte; avere partecipato a conferenze dove ho essenzialmente parlato con sconosciuti o di fronte ai guru degli argomenti che stavo studiando; di avere trascorso alcuni mesi all'estero, da sola, lontana da casa.E, perchè no, ci tengo anche a ricordare di tutte le belle opportunità che ho avuto di vedere nuovi posti nel mondo: dalla mia adorata Siviglia, a Lisbona, alla Bassa Sassonia, a Cipro e Malta e, non da ultimo, alla California.Chi mai avrebbe pensato di andare anche in America?! Penso e spero che queste righe abbiano dato un quadro, seppur non esaustivo, di ciò che per me sono stati questi 3 anni.Ma niente di tutto quello che ho citato sarebbe stato possibile senza tutte le persone che mi hanno accompagnato in questo bel percorso (e che, almeno per ora, non si libereranno di me), spronandomi sempre in tutte le attività.Ci tengo quindi a ringraziare tutto il mio gruppo di ricerca.Grazie ai miei supervisor, Maurizio Mongelli e Fabrizio Dabbene, per la loro importante guida scientifica, che mi hanno dato sempre le giuste motivazioni e permesso di avere tante belle opportunità.Grazie ai miei compagni di avventura nonchè preziosi collaboratori: grazie ad Alberto, che mi ha aiutata a fare un pochino più mia la matematica e mi ha sempre supportato quando più ne avevo bisogno; grazie a Marta, con cui ho condiviso tante vi belle esperienze, soprattutto nella nostra \"terra d'origine\" dell'ingegneria biomedica.Grazie anche ai collaboratori \"passati\", Giacomo, Ivan e Vanessa, che hanno contribuito con me ad alcuni lavori che oggi presento in questa tesi.E grazie infine a tutti gli altri colleghi di Genova, \"quelli dei pranzi belli': Alessandra, Barbara, Enrico (a cui va la menzione speciale per avermi avviato al python e no, non rinnego il passato cyber), Martina, Sabrina, Sandro, Paola, con cui le risate non mancano mai.Non posso poi non ringraziare la mia famiglia: i miei genitori, che mi hanno aiutato tanto in questi anni, permettendomi di essere giunta a questo importante traguardo; mia sorella Elena, che -seppur a suo modo -sa sempre come rincuorarmi e che, finalmente, adesso non mi sentirà più dire \"I'm Sara Narteni, a PhD student,...\" mentre provo le presentazioni.Grazie ai miei nonni, Anna, Armando e Mariuccia, che per me ci sono sempre, anche se per loro il funzionamento di un dottorato rimarrà un mistero maggiore di quelli che si nascondono dietro all'Intelligenza Artificiale ;-).Non so bene cosa mi riserverà il futuro, ma... Ad Maiora!Sara vii \"La vita non è facile per nessuno di noi.Ma cosa importa?Bisogna perseverare e, soprattutto, avere fiducia in sé stessi.Bisogna credere di essere dotati per qualcosa e che questa cosa deve essere raggiunta.\"M. Curie Alla mia famiglia.A.2 Examples from REXASI-PRO safety analysis.Each spider chart shows how different combinations of values for the ODD variables (reported in bold) determine a different ASIL level of risk for the individuated malfunctioning behaviors and hazards (QM: quality management; ASIL-A: low risk; ASIL-B: moderate risk; ASIL-C: high risk).ASIL is in turn assigned from exposure, severity and controllability scores.Risk levels other than QM require the introduction of safety goals to mitigate them. . . . . . . . . . . . . . . . . . . . .124 A.3 Schematic summary of the rule-based OoD detection approach. . . .126 A.4 Visual intuition behind the need for deep probabilistic scaling approach.The goal is to understand which samples are \"not conformal\" with respect to a target class. . . . . . . . . . . . . . . . . . . . . .128 A.5 Distribution of uncertainty of MNIST17 and PneumoniaMNIST.Shown in blue and yellow, respectively, are the uncertain classifications for class '1' and class '7' for MNIST1-7 and for class \"normal\" and class \"pneumonia\" for PneumoniaMNIST. . . . . . . . . . . . . . . . .", "source_url": "https:\/\/openalex.org\/W7130587133", "pdf_url": "https:\/\/iris.polito.it\/bitstream\/11583\/2999512\/2\/conv_phd_thesis_final_20250218.pdf", "doi": "10.13121\/polito\/porto\/2999512", "handle": "", "repository_name": "PORTO@Iris", "source_record_id": "W7130587133", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1173, "title": "Causal Structure Learning for Robust and Explainable Artificial Intelligence", "title_ar": "", "title_en": "Causal Structure Learning for Robust and Explainable Artificial Intelligence", "author": "Brandon David Mossop", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Queen's University Library", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; nlp; responsible-ai; predictive-analytics; llm; xai", "keywords": "", "abstract": "Modern artificial intelligence (AI) has achieved success by learning statistical patterns from data and framing many problems as prediction tasks. However, prediction alone cannot answer questions such as why an outcome occurred or how an intervention would change it. Answering these questions requires reasoning about cause and effect. Causal AI addresses this limitation by incorporating causal inference into AI. This thesis advances that goal through causal structure learning, focusing on two interrelated challenges: causal forecasting and causal discovery. First, it demonstrates that causal feature selection can improve both the predictive performance and explainability of AI forecasting. It then extends this idea to non-stationary environments by modeling regime-dependent causal structures, providing improved forecasts and interpretable insights into how underlying mechanisms evolve over time. Because these forecasting approaches rely on causal structure, the thesis next turns to causal discovery. It develops an agentic framework using open-weight small language models that combines statistical evidence with agents for orienting, augmenting, and critiquing causal graphs. The framework is then extended with a causal story agent that reasons about the data-generating process to construct more accurate causal graphs. Collectively, this thesis advances causal AI toward systems that represent the mechanisms underlying observed outcomes rather than only statistical associations.", "source_url": "https:\/\/openalex.org\/W7223075870", "pdf_url": "https:\/\/queensu.scholaris.ca\/bitstreams\/f2f2be29-1185-4c08-9071-57eaf93ac9da\/download", "doi": "", "handle": "", "repository_name": "QSpace (Queen's University Library)", "source_record_id": "W7223075870", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1178, "title": "Human interpretable artificial intelligence applications for microbial-related diseases", "title_ar": "", "title_en": "Human interpretable artificial intelligence applications for microbial-related diseases", "author": "Josh L. Espinoza", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Durban University of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "South Africa", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "medical-ai", "research_subdomain": "xai", "categories": "machine-learning; medical-ai; responsible-ai; ai-applications; xai; ai-education", "keywords": "", "abstract": "The human microbiome is a complex ecosystem that is influenced not only by host genetics but environmental stimuli. With advancements in next-generation sequencing (NGS) technologies, genomics and related meta-omics such as metagenomics, metatranscriptomics and metaproteomics have become increasingly accessible for researchers and clinicians to investigate microbial-related diseases. However, analysis of the outputs of “omics” technologies are often difficult due to variance introduced by biological complexity, batch effects from laboratory protocols\/conditions, and the sensitivity\/calibration of highly sensitive instruments. The biological complexity of “omics” presents a considerable analytical obstacle as most datasets contain hundreds of thousands to millions of unique features with unknown connections and nested hierarchies. In addition to this inherent complexity, the deluge of data generated from NGS technologies is fundamentally compositional, conveys only relative information, and because of this cannot be robustly analyzed using conventional statistical approaches. Furthermore, meta-omics datasets are typically sparse and the number of biological features often vastly exceeds the number of biological samples which can introduce anomalies in statistical analysis and the downstream findings if not addressed accordingly; a term dubbed as “the curse of dimensionality”. The complexity, compositionality, and dimensionality of “omics” datasets makes it challenging to derive clinical meaning and an understanding of the microbial system with respect to a host phenotype. Although, artificial intelligence and machine-learning methods have progressed substantially in recent years, their applications in domain sciences such as biology, and by extension “omics” technologies, have been limited in terms of human interpretability. In many machine-learning paradigms, interpretability is often sacrificed for analytical performance, or vice versa, but recently a domain-agnostic effort aims to develop explainable artificial intelligence algorithms that have both high modeling performance and human interpretability; a major goal of biomedical sciences. In this dissertation, I develop novel approaches in bridging biological science with machine learning methods at the vanguard of scientific development through the initiative of explainable artificial intelligence. The methods developed are validated on 3 datasets pertaining to microbial-related diseases including antibiotic resistance discovery, acute malnutrition in West African children, and caries pathology in Australian juvenile twins. The combination of methods developed are expected to provide the means for clinical researchers to overcome obstacles in interrogating the complex narratives that determine health and disease.", "source_url": "https:\/\/openalex.org\/W4362630824", "pdf_url": "https:\/\/openscholar.dut.ac.za\/bitstream\/10321\/4701\/1\/Espinoza%20JL%20Final%20Dissertation_Redacted.pdf", "doi": "10.51415\/10321\/4701", "handle": "", "repository_name": "DUT Open Scholar (Durban University of Technology)", "source_record_id": "W4362630824", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1181, "title": "Explainable artificial intelligence for the electrocardiogram", "title_ar": "", "title_en": "Explainable artificial intelligence for the electrocardiogram", "author": "Rutger R van de Leur", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Utrecht University", "university_source": "repository_host", "college": "", "department": "", "country": "Netherlands", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "medical-ai", "research_subdomain": "xai", "categories": "machine-learning; deep-learning; medical-ai; responsible-ai; predictive-analytics; xai; fairness", "keywords": "", "abstract": "The 12-lead electrocardiogram (ECG) is a crucial diagnostic tool in clinical practice, used to identify various cardiac abnormalities. However, interpreting ECGs is complex, with significant variability between and within clinicians, and expert knowledge is not always accessible. This has led to the development of computerized interpretation algorithms (CIE), though they have not yet matched the accuracy of physicians. Recent advancements, particularly the use of deep neural networks (DNNs), have shown potential to process raw ECG data more effectively without relying on hand-crafted features. With the availability of large labeled ECG datasets, ECGs are ideal for developing AI-based algorithms. This research aimed to bridge the gap between AI advancements and their clinical application, bringing these technologies closer to real-world use. Four key prerequisites for developing and implementing ECG-AI were identified: asking the right clinical questions, ensuring rigorous quality control, addressing potential bias using explainability techniques, and conducting appropriate implementation studies. Additionally, we highlighted four key opportunities for ECG-AI applications: disease screening, diagnostic workflow optimization, decision support in treatment, and detecting new ECG features. The first step in exploring DNNs for ECG interpretation involved developing an algorithm to optimize diagnostic workflows. This algorithm efficiently triaged ECGs, identifying which required expert consultation and when, ensuring safe and effective clinical outcomes. An implementation study in a hospital setting demonstrated the algorithm’s practical benefits. Another important aspect of AI implementation is ensuring the algorithms can express uncertainty, as DNNs tend to make predictions even with noisy or unfamiliar data. We developed a method to quantify the algorithm's confidence in its predictions, ensuring that only certain ECGs undergo automatic analysis, while others are referred to an expert. Addressing challenges in DNN usage, including the lack of explainability and the need for large datasets, we created a novel method to interpret ECGs in an explainable manner. By training a variational auto-encoder on over one million ECGs, we decomposed the ECG morphology into 32 explainable factors, termed FactorECG. This method performed as effectively as traditional \"black box\" models but with added transparency, offering promise in applications like detecting reduced ejection fraction. FactorECG also surpassed heatmap-based techniques in improving explainability. Given that large ECG datasets are often unavailable for certain clinical questions, we explored the transferability of DNN knowledge from large to small datasets. We developed algorithms to detect specific genetic mutations and predict disease progression in patients with various heart conditions. Remarkably, these algorithms outperformed traditional prediction models, relying only on the 12-lead ECG for input. In conclusion, while significant progress has been made in developing ECG-AI, challenges remain before it can be fully implemented in clinical practice. Future work must focus on refining the FactorECG algorithm, conducting large-scale trials, and developing software platforms to integrate these algorithms into clinical workflows. Importantly, further research is needed to assess whether ECG-AI improves patient outcomes and reduces healthcare costs and burdens.", "source_url": "https:\/\/openalex.org\/W4402589842", "pdf_url": "", "doi": "10.33540\/2459", "handle": "", "repository_name": "Utrecht University Repository (Utrecht University)", "source_record_id": "W4402589842", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1185, "title": "Machine Learning for Predictive Maintenance in Aviation", "title_ar": "", "title_en": "", "author": "Panagiotis Korvesis", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2017, "language": "fr", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "machine-learning; deep-learning; computer-vision; nlp; ai-applications; predictive-analytics; speech-recognition; ai-industry", "keywords": "", "abstract": "Apprentissage Automatique pour la Maintenance Predictive dans le Domaine de l’Aviation L'augmentation des données disponibles dans presque tous les domaines soulève la nécessité d'utiliser des algorithmes pour l'analyse automatisée des données. Cette nécessité est mise en évidence dans la maintenance prédictive, où l'objectif est de prédire les pannes des systèmes en observant continuellement leur état, afin de planifier les actions de maintenance à l'avance. Ces observations sont générées par des systèmes de surveillance habituellement sous la forme de séries temporelles et de journaux d'événements et couvrent la durée de vie des composants correspondants. Le principal défi de la maintenance prédictive est l'analyse de l'historique d'observation afin de développer des modèles prédictifs.Dans ce sens, l'apprentissage automatique est devenu omniprésent puisqu'il fournit les moyens d'extraire les connaissances d'une grande variété de sources de données avec une intervention humaine minimale. L'objectif de cette thèse est d'étudier et de résoudre les problèmes dans l'aviation liés à la prévision des pannes de composants à bord. La quantité de données liées à l'exploitation des avions est énorme et, par conséquent, l'évolutivité est une condition essentielle dans chaque approche proposée.Cette thèse est divisée en trois parties qui correspondent aux différentes sources de données que nous avons rencontrées au cours de notre travail. Dans la première partie, nous avons ciblé le problème de la prédiction des pannes des systèmes, compte tenu de l'historique des Post Flight Reports. Nous avons proposé une approche statistique basée sur la régression précédée d'une formulation méticuleuse et d'un prétraitement \/ transformation de données. Notre méthode estime le risque d'échec avec une solution évolutive, déployée dans un environnement de cluster en apprentissage et en déploiement. À notre connaissance, il n'y a pas de méthode disponible pour résoudre ce problème jusqu'au moment où cette thèse a été écrite.La deuxième partie consiste à analyser les données du livre de bord, qui consistent en un texte décrivant les problèmes d'avions et les actions de maintenance correspondantes. Le livre de bord contient des informations qui ne sont pas présentes dans les Post Flight Reports bien qu'elles soient essentielles dans plusieurs applications, comme la prédiction de l'échec. Cependant, le journal de bord contient du texte écrit par des humains, il contient beaucoup de bruit qui doit être supprimé afin d'extraire les informations utiles. Nous avons abordé ce problème en proposant une approche basée sur des représentations vectorielles de mots. Notre approche exploite des similitudes sémantiques, apprises par des neural networks qui ont généré les représentations vectorielles, afin d'identifier et de corriger les fautes d'orthographe et les abréviations. Enfin, des mots-clés importants sont extraits à l'aide du Part of Speech Tagging.Dans la troisième partie, nous avons abordé le problème de l'évaluation de l'état des composants à bord en utilisant les mesures des capteurs. Dans les cas considérés, l'état du composant est évalué par l'ampleur de la fluctuation du capteur et une tendance à l'augmentation monotone. Dans notre approche, nous avons formulé un problème de décomposition des séries temporelles afin de séparer les fluctuations de la tendance en résolvant un problème convexe. Pour quantifier l'état du composant, nous calculons à l'aide de Gaussian Mixture Models une fonction de risque qui mesure l'écart du capteur par rapport à son comportement normal.", "source_url": "https:\/\/openalex.org\/W7148425078", "pdf_url": "https:\/\/pastel.hal.science\/tel-02003508v1\/file\/65181_KORVESIS_2017_archivage.pdf", "doi": "10.70675\/5c156250zffd4z4dc8z8067zd9361665aa02", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7148425078", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1186, "title": "Predictive Maintenance for IT Equipment using Machine Learning Models", "title_ar": "", "title_en": "Predictive Maintenance for IT Equipment using Machine Learning Models", "author": "Rashed Khalid Alshirawi", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Rochester Institute of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "responsible-ai", "research_subdomain": "xai", "categories": "machine-learning; responsible-ai; ai-applications; predictive-analytics; xai; ai-industry", "keywords": "", "abstract": "Predictive maintenance was recently introduced as a paradigm shift in the field of smart manufacturing, and it allows organizations to predict equipment failures and streamline maintenance planning based on the data-driven insights. The present work addresses the concept of the implementation of a decision tree-based predictive maintenance model based on AI4I 2020 Predictive Maintenance Dataset, a simulated environment of an actual industry through sensor data including temperature, torque, rotational velocity, and toolwear. The study uses a quantitative approach that is based on positivism philosophy, which is the focus on objectivity, empirical validation, and statistical testing of the model performance. Synthetic Minority Oversampling Technique (SMOTE) was used to solve the problem of skewed data in the dataset wherein the distribution has been equalized between the failure and non-failure cases. This preprocessing procedure raised the ability of the model to identify minority-class events, such as machine failures, and this negatively affected the recall and the overall classification accuracy. The Decision Tree classifier was obtained with the accuracy of 90.76 percentage backed up by the precision, recall, and F1-score values that show good prediction performance and at the same time, interpretability which is imperative in any industrial use. The feature-importance analysis showed that the variables concerning Heat Dissipation Failure, Power Failure, and Overstrain Failure were the most significant variables that predict equipment breakdowns. The results of the study indicate that the combination of machinelearning interpretability and balanced and high-quality data is essential in the process of obtaining reliable predictive maintenance solutions. The findings confirm the Decision Tree as a realistic and explicable decision model to use in the practical application and provides transparency as well as accuracy in the maintenance decision making. The study culminates in suggestions on how continuous monitoring can be introduced, sensor networks can be extended, and real time earlywarning dashboards can be developed to make industries efficient and resilient to operations. The next steps inwork are to investigate the techniques of ensemble learning and hybrid predictive systems to enhance further the accuracy of fault-detection and adaptability to a tough manufacturing environment.", "source_url": "https:\/\/openalex.org\/W7119561594", "pdf_url": "", "doi": "", "handle": "", "repository_name": "RIT Scholar Works (Rochester Institute of Technology)", "source_record_id": "W7119561594", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1192, "title": "Machine Learning with Conformal Prediction for Predictive Maintenance tasks in Industry 4.0 : Data-driven Approach", "title_ar": "", "title_en": "Machine Learning with Conformal Prediction for Predictive Maintenance tasks in Industry 4.0 : Data-driven Approach", "author": "Liu, Shuzhou ; Mulahuko, Mpova", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Jönköping University", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "deep-learning", "research_subdomain": "rnn", "categories": "machine-learning; deep-learning; ai-applications; predictive-analytics; rnn; ai-industry", "keywords": "", "abstract": "This thesis is a cooperation with Knowit, Östrand \\& Hansen, and Orkla. It aimed to explore the application of Machine Learning and Deep Learning models with Conformal Prediction for a predictive maintenance situation at Orkla. Predictive maintenance is essential in numerous industrial manufacturing scenarios. It can help to reduce machine downtime, improve equipment reliability, and save unnecessary costs. In this thesis, various Machine Learning and Deep Learning models, including Decision Tree, Random Forest, Support Vector Regression, Gradient Boosting, and Long short-term memory, are applied to a real-world predictive maintenance dataset. The Orkla dataset was originally planned to use in this thesis project. However, due to some challenges met and time limitations, one NASA C-MAPSS dataset with a similar data structure was chosen to study how Machine Learning models could be applied to predict the remaining useful lifetime (RUL) in manufacturing. Besides, conformal prediction, a recently developed framework to measure the prediction uncertainty of Machine Learning models, is also integrated into the models for more reliable RUL prediction. The thesis project results show that both the Machine Learning and Deep Learning models with conformal prediction could predict RUL closer to the true RUL while LSTM outperforms the Machine Learning models. Also, the conformal prediction intervals provide informative and reliable information about the uncertainty of the predictions, which can help inform personnel at factories in advance to take necessary maintenance actions. Overall, this thesis demonstrates the effectiveness of utilizing machine learning and Deep Learning models with Conformal Prediction for predictive maintenance situations. Moreover, based on the modeling results of the NASA dataset, some insights are discussed on how to transfer these experiences into Orkla data for RUL prediction in the future.", "source_url": "https:\/\/openalex.org\/W7113257904", "pdf_url": "https:\/\/hj.diva-portal.org\/smash\/get\/diva2:1765779\/FULLTEXT01", "doi": "", "handle": "", "repository_name": "Jonkoping University Library (Jönköping University)", "source_record_id": "W7113257904", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1193, "title": "Application of Machine Learning Methods on Predictive Maintenance", "title_ar": "", "title_en": "Application of Machine Learning Methods on Predictive Maintenance", "author": "Feranmi Akanni", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Tampere University", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "machine-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; ai-applications; predictive-analytics; ai-industry", "keywords": "", "abstract": "Maintenance is important in optimization of the business value of a functional unit and this optimization can only be achieved through predictive maintenance which ensures that maintenance activities are not carried out before due time and at the same time, prevent occurrence of breakdown of functional units because of missed maintenance activities.In this thesis, we focus on using different machine learning methods to predict the failure of a functional units.We explore the data and use missing data techniques to deal with missing values in the dataset, which resulted in a complete dataset.We explore various feature selection techniques to extract important features and reduce dimensionality of the dataset.Then, we explore the following machine learning methods: logistic regression, naïve Bayes, support vector machine, k-nearest neighbour and ensemble learning techniques which are bagging and boosting methods.Our results indicated that predictions from ensemble learning techniques have better evaluation metrics compared to other machine learning methods.", "source_url": "https:\/\/openalex.org\/W3010495304", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Tampere University Institutional Repository (Tampere University)", "source_record_id": "W3010495304", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1194, "title": "Predictive Maintenance Using Machine Learning Methods in Petrochemical Refineries", "title_ar": "", "title_en": "Predictive Maintenance Using Machine Learning Methods in Petrochemical Refineries", "author": "Sander Suursalu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Delft University of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Netherlands", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "deep-learning", "research_subdomain": "rnn", "categories": "machine-learning; deep-learning; ai-applications; predictive-analytics; rnn; ai-industry", "keywords": "", "abstract": "This research project evaluates the suitability of machine learning methods for early fault prediction and predictive maintenance in petrochemical refineries based on real- life use cases at Shell Pernis. Refineries are mature industrial installations, however, unplanned shutdowns still occur due to equipment failures. Refineries have petabytes of process control data available from the past years, however, all of that data is unla- belled. The goal of this research project was to evaluate, whether useful information can be extracted from the process control data. The resulting approach had to be compatible with Shell IT, scalable to larger sections of the refinery, reusable in other parts of the refinery and capable of detecting the components that cause the potential faults. During this research project, multiple solutions based on artificial neural net- works and statistical approaches were implemented to model the normal behaviour of the monitored systems. Abnormal predictions for the modelled systems were then used to predict failures in advance, where the prediction horizon reached more than a month for some use cases. 4-layer GRUs with tanh activation functions and an input sequence length of 4 samples provided the best results. GRUs were 7% faster to train than LSTMs while reducing the prediction error by 15%. Furthermore, the predic- tion error was less than 3% for the normal operating conditions while reaching more than 15% prior to failures. Therefore, machine learning models can predict failures in petrochemical refineries without any industry-specific knowledge, if the model is trained with clean data that does not contain abnormal time series.", "source_url": "https:\/\/openalex.org\/W2779737813", "pdf_url": "https:\/\/repository.tudelft.nl\/file\/File_fa0ac32d-b9c9-49f2-b716-3d1fe9d9008f", "doi": "", "handle": "", "repository_name": "Research Repository (Delft University of Technology)", "source_record_id": "W2779737813", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1198, "title": "Predictive Maintenance Using Machine Learning", "title_ar": "", "title_en": "", "author": "Mateyaunga Innocent", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Algeria", "governorate": "", "publication_year": 2020, "language": "fr", "research_domain": "deep-learning", "research_subdomain": "rnn", "categories": "machine-learning; deep-learning; ai-applications; predictive-analytics; rnn; ai-industry", "keywords": "", "abstract": "Ces dernières années, le Deep Learning a attiré une attention croissante, d’où son application dans la maintenance prédictive. Il s’agit d’un domaine de recherche en évolution qui a le potentiel d’influencer divers domaines d’application et, par conséquent, son utilisation pour les applications de gestion de la santé des systèmes et équipements complexes et couteux doit être mis à profit. Cette thèse présente une revue systématique de la gestion de la santé du système basée sur l’intelligence artificielle (IA) en mettant l’accent sur les tendances récentes de l’apprentissage profond (DL) dans le domaine: différents types de modèles d’apprentissage automatique (ML) qui peuvent être utilisés pour prédire la défaillance d’une machine .Au début, la revue des publications précédentes est effectué, puis différents modèles (LSTM, GRU et SRN) sont utilisés sur l’ensemble de données NASA Turbofan afin de déterminer le meilleur modèle, les versions CuDNN (CuDNNLSTM, CuDNNGRU) sont également appliquées et le temps d’exécution est comparé aux temps de base (LSTM, GRU). Le CuDNNGRU s’avère être la meilleure méthode en raison de son temps d’exécution (30,6 s) et de sa exactitude compétitive (0,9827) et de sa précision (0,9452).", "source_url": "https:\/\/openalex.org\/W7135298816", "pdf_url": "https:\/\/dspace.univ-tlemcen.dz\/bitstreams\/3dec6713-7835-4c54-92a5-a7922e1ca84a\/download", "doi": "", "handle": "", "repository_name": "Dépôt Institutionnel de lUniversité de Tlemcen", "source_record_id": "W7135298816", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1205, "title": "Predictive maintenance with machine learning on weld joint analysed by ultrasound", "title_ar": "", "title_en": "Predictive maintenance with machine learning on weld joint analysed by ultrasound", "author": "Adam Hedkvist", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Uppsala University", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; ai-applications; predictive-analytics; cnn; medical-imaging; ai-industry", "keywords": "", "abstract": "Ever since the first industrial revolution industries have had the goal to increase their production. With new technology such as CPS, AI and IoT industries today are going through the fourth industrial revolution denoted as industry 4.0. The new technology not only revolutionises production, but also maintenance, making predictive maintenance possible. Predictive maintenance seeks to predict when failure would occur, instead of having scheduled maintenance or maintenance after failure already occurred. In this report a convolutional neural network (CNN) will analyse data from an ultrasound machine scanning a weld joint. The data from the ultrasound machine will be transformed by the short time Fourier transform in order to create an image for the CNN. Since the data from the ultrasound is not complete, simulated data will be created and investigated as another option for training the network. The results are promising, however the lack of data makes it hard to show any concrete proof.", "source_url": "https:\/\/openalex.org\/W2988297880", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Uppsala University Publications (Uppsala University)", "source_record_id": "W2988297880", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1208, "title": "أثرمنهج التعلم العميق القائم على الذكاء الاصطناعي على مهارة كتابة اللغة العربية لدى طلاب الصف الحادي عشر في المدرسة الثانوية الحكومية الثانية بيوجياكرتا ٢٠٢٦ = ATHR MANHAJ AL-TA'ALLUM AL-'AMIQ AL-QA'IM 'ALA AL-DHAKA' AL-ISTINA'I 'ALA MAHARAT KITABAT AL-LUGHAH AL-'ARABIYYAH LADA TULLAB AL-SAFF AL-HADI 'ASHAR FI AL-MADRASAH AL-THANAWIYYAH AL-HUKUMIYYAH AL-THANIYAH YOGYAKARTA 2026", "title_ar": "أثرمنهج التعلم العميق القائم على الذكاء الاصطناعي على مهارة كتابة اللغة العربية لدى طلاب الصف الحادي عشر في المدرسة الثانوية الحكومية الثانية بيوجياكرتا ٢٠٢٦ = ATHR MANHAJ AL-TA'ALLUM AL-'AMIQ AL-QA'IM 'ALA AL-DHAKA' AL-ISTINA'I 'ALA MAHARAT KITABAT AL-LUGHAH AL-'ARABIYYAH LADA TULLAB AL-SAFF AL-HADI 'ASHAR FI AL-MADRASAH AL-THANAWIYYAH AL-HUKUMIYYAH AL-THANIYAH YOGYAKARTA 2026", "title_en": "", "author": "NIM.: 22104020014 Sri Fatimah", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Sunan Kalijaga State Islamic University", "university_source": "repository_host", "college": "", "department": "", "country": "Indonesia", "governorate": "", "publication_year": 2026, "language": "ar", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "deep-learning; nlp; transformers; arabic-nlp", "keywords": "", "abstract": "Penelitian ini dilatarbelakangi oleh masih rendahnya keterampilan menulis Bahasa Arab siswa kelas XI MAN 2 Yogyakarta, khususnya dalam menulis kalimat dan paragraf sederhana. Penelitian ini bertujuan untuk mengetahui pengaruh pendekatan deep learning berbasis Artificial Intelligence (AI) terhadap keterampilan menulis Bahasa Arab siswa kelas XI MAN 2 Yogyakarta. Penelitian ini menggunakan pendekatan kuantitatif dengan jenis kuasi-eksperimen dan desain non-equivalent control group design. Populasi penelitian adalah seluruh siswa kelas XI MAN 2 Yogyakarta tahun ajaran 2025\/2026 dengan teknik pengambilan cluster random sampling. Pengumpulan data dilakukan melalui pretest, posttest, dan observasi partisipan. Analisis data menggunakan uji Wilcoxon, uji Mann-Whitney U, dan uji N-Gain. Hasil penelitian menunjukkan bahwa pendekatan deep learning berbasis AI tidak berpengaruh secara signifikan terhadap keterampilan menulis Bahasa Arab siswa kelas XI MAN 2 Yogyakarta. Hasil uji Mann-Whitney U menunjukkan nilai signifikansi sebesar 0,221 pada analisis pertama dan 0,133 pada analisis kedua, yang berarti lebih besar dari 0,05. Selain itu, hasil uji N-Gain memperoleh nilai sebesar 0,041 yang termasuk kategori rendah. Dengan demikian, pendekatan deep learning berbasis AI hanya menunjukkan peningkatan yang rendah dan tidak signifikan secara statistik terhadap keterampilan menulis Bahasa Arab siswa.", "source_url": "https:\/\/openalex.org\/W7164774459", "pdf_url": "https:\/\/digilib.uin-suka.ac.id\/id\/eprint\/76688\/1\/22104020014_BAB-I_IV-atau-V_DAFTAR-PUSTAKA.pdf", "doi": "", "handle": "", "repository_name": "Digital Library UIN Sunan Kalijaga (Sunan Kalijaga State Islamic University)", "source_record_id": "W7164774459", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1209, "title": "Pengaruh artificial intelligence self-eficacy terhadap stres akademik mahasiswa Fakultas Psikologi Universitas Islam Negeri Maulana Malik Ibrahim Malang", "title_ar": "", "title_en": "", "author": "Ahmad Rofiul Abror", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Maulana Malik Ibrahim State Islamic University", "university_source": "repository_host", "college": "", "department": "", "country": "Indonesia", "governorate": "", "publication_year": 2025, "language": "id", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; nlp; ai-applications; predictive-analytics; transformers; arabic-nlp; ai-education", "keywords": "", "abstract": "INDONESIA Penelitian ini bertujuan untuk mengetahui pengaruh Artificial Intelligence Self-Efficacy (AISE) terhadap stres akademik pada mahasiswa Fakultas Psikologi UIN Malang. Latar belakang penelitian ini didasari oleh meningkatnya penggunaan kecerdasan buatan (AI) dalam aktivitas akademik mahasiswa serta asumsi bahwa tingkat kepercayaan diri terhadap penggunaan AI dapat memengaruhi kondisi psikologis, khususnya stres akademik. Metode penelitian yang digunakan adalah pendekatan kuantitatif dengan teknik analisis regresi linier sederhana. Sampel penelitian terdiri dari mahasiswa aktif Fakultas Psikologi UIN Maulana Malik Ibrahim Malang yang dipilih melalui teknik purposive sampling dengan kriteria: terdaftar sebagai mahasiswa aktif pada semester berjalan dan pernah menggunakan teknologi berbasis Artificial Intelligence (AI) dalam aktivitas akademik seperti menyusun tugas, mencari referensi, atau membuat laporan. Hasil penelitian menunjukkan bahwa AISE tidak berpengaruh signifikan terhadap stres akademik, dengan nilai koefisien regresi sebesar -0,182 dan signifikansi 0,098 (p > 0,05). Temuan ini menunjukkan bahwa kepercayaan diri mahasiswa dalam menggunakan teknologi AI belum cukup untuk mengurangi tingkat stres akademik. Implikasi dari hasil ini adalah perlunya pengembangan program literasi dan pelatihan penggunaan AI yang lebih komprehensif, agar mahasiswa tidak hanya mampu memanfaatkan teknologi tersebut secara teknis, tetapi juga dapat menggunakannya secara adaptif dan efektif dalam menghadapi tuntutan akademik tanpa menimbulkan tekanan psikologis. Faktor lain seperti dukungan sosial, beban akademik, dan kecemasan terhadap penggunaan AI diduga lebih berperan dalam memengaruhi stres. Penelitian ini diharapkan dapat memberikan pemahaman baru mengenai pentingnya keseimbangan antara kemampuan teknologis dan kesiapan psikologis mahasiswa di era digital. ENGLISH This study aims to determine the effect of Artificial Intelligence Self-Efficacy (AISE) on academic stress in students of the Faculty of Psychology, UIN Malang. The background of this study is based on the increasing use of artificial intelligence (AI) in students' academic activities and the assumption that the level of confidence in the use of AI can affect psychological conditions, especially academic stress. The research method used is a quantitative approach with a simple linear regression analysis technique. The research sample consisted of active students of the Faculty of Psychology, UIN Maulana Malik Ibrahim Malang who were selected through a purposive sampling technique with the following criteria: registered as active students in the current semester and have used Artificial Intelligence (AI)-based technology in academic activities such as compiling assignments, searching for references, or making reports. The results showed that AISE had no significant effect on academic stress, with a regression coefficient value of -0.182 and a significance of 0.098 (p > 0.05). This finding indicates that students' confidence in using AI technology is not enough to reduce academic stress levels. The implication of these results is the need to develop more comprehensive AI literacy and training programs, so that students are not only able to utilize the technology technically, but also can use it adaptively and effectively to face academic demands without causing psychological stress. Other factors such as social support, academic burden, and anxiety about AI use are suspected to play a larger role in influencing stress. This research is expected to provide new insights into the importance of balancing technological skills and psychological readiness in students in the digital era. ARABIC تهدف هذه الدراسة إلى تحديد أثر الكفاءة الذاتية في استخدام الذكاء الاصطناعي على الضغط الأكاديمي لدى طلاب كلية علم النفس بجامعة مالانج الإسلامية الحكومية. وتستند خلفية هذه الدراسة إلى تزايد استخدام الذكاء الاصطناعي في الأنشطة الأكاديمية للطلاب، وفرضية أن مستوى الثقة في استخدام الذكاء الاصطناعي قد يؤثر على الحالة النفسية، وخاصة الضغط الأكاديمي. استخدمت الدراسة المنهج الكمي، وتحديدًا تقنية تحليل الانحدار الخطي البسيط. تكونت عينة البحث من طلاب كلية علم النفس بجامعة مولانا مالك إبراهيم الإسلامية الحكومية في مالانج، والذين تم اختيارهم باستخدام أسلوب العينة الهادفة وفقًا للمعايير التالية: أن يكونوا مسجلين كطلاب منتظمين في الفصل الدراسي الحالي، وأن يكونوا قد استخدموا تقنية الذكاء الاصطناعي في أنشطة أكاديمية مثل إعداد الواجبات، والبحث عن المراجع، وكتابة التقارير. أظهرت النتائج أن الكفاءة الذاتية في استخدام الذكاء الاصطناعي لم يكن لها تأثير معنوي على الضغط الأكاديمي، حيث بلغ معامل الانحدار -0.182، ومستوى الدلالة 0.098 (قيمة الاحتمال > 0.05). تشير هذه النتيجة إلى أن ثقة الطلاب في استخدام تقنيات الذكاء الاصطناعي لا تكفي وحدها لخفض مستويات التوتر الأكاديمي.", "source_url": "https:\/\/openalex.org\/W7125361256", "pdf_url": "", "doi": "", "handle": "", "repository_name": "eTheses of Maulana Malik Ibrahim State Islamic University (Maulana Malik Ibrahim State Islamic University)", "source_record_id": "W7125361256", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1211, "title": "Students' perception of using Chat Gpt as an academic writing assistant at an Islamic Senior High School", "title_ar": "", "title_en": "Students' perception of using Chat Gpt as an academic writing assistant at an Islamic Senior High School", "author": "Bunayya Izzani Fathi Musaqqaf", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Maulana Malik Ibrahim State Islamic University", "university_source": "repository_host", "college": "", "department": "", "country": "Indonesia", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "nlp; responsible-ai; ai-applications; llm; fairness; ai-education", "keywords": "", "abstract": "INGGRIS The growing phenomenon of Artificial Intelligence (AI) use, particularly ChatGPT, in academic writing has become an interesting topic recently. While ChatGPT assists students in generating ideas, improving sentence structure, and enhancing writing efficiency, its use also raises concerns about dependency, ethics, and declining critical thinking skills. The purpose of this research is to explore students’ perceptions of using ChatGPT in academic writing and to identify the challenges they experience. This study employed a qualitative approach with a transcendental phenomenological design, where data were collected through observation and semi structured interviews involving 23 students from the 11th grade “Excellent (A)” class at MAN 1 Nganjuk, six of whom were selected as key informants. Data collection was carried out through direct classroom observation during writing activities and interviews to understand students’ experiences and viewpoints. The findings reveal that students hold a positive perception of ChatGPT as a helpful tool that supports idea generation and writing organization. However, several technical challenges such as unstable internet connections and cognitive challenges including poor prompt literacy, dependence on AI outputs, and reduced creativity were also identified. The discussion highlights that the effectiveness of ChatGPT in academic writing depends largely on students’ learning motivation, digital literacy, and teachers’ guidance in fostering ethical and reflective AI use. INDONESIA Fenomena meningkatnya penggunaan Artificial Intelligence (AI), khususnya ChatGPT, dalam penulisan akademik menjadi topik yang menarik perhatian dalam beberapa waktu terakhir. ChatGPT membantu siswa dalam menghasilkan ide, memperbaiki struktur kalimat, dan meningkatkan efisiensi penulisan. Namun, penggunaannya juga memunculkan kekhawatiran terkait ketergantungan, etika, serta menurunnya kemampuan berpikir kritis. Tujuan penelitian ini adalah untuk mengeksplorasi bagaimana persepsi siswa terhadap penggunaan ChatGPT dalam penulisan akademik serta mengidentifikasi tantangan yang mereka alami. Penelitian ini menggunakan pendekatan kualitatif dengan desain fenomenologi transendental. Data dikumpulkan melalui observasi dan wawancara semi terstruktur terhadap 23 siswa kelas XI “Unggulan (A)” di MAN 1 Nganjuk, dengan enam siswa dipilih sebagai informan kunci. Pengumpulan data dilakukan melalui observasi langsung selama kegiatan menulis dan wawancara untuk memahami pengalaman serta pandangan siswa. Hasil penelitian menunjukkan bahwa siswa memiliki persepsi positif terhadap ChatGPT sebagai alat bantu yang mendukung pengembangan ide dan pengorganisasian tulisan. Namun, beberapa tantangan juga ditemukan, seperti kendala teknis berupa koneksi internet yang tidak stabil, serta tantangan kognitif seperti rendahnya kemampuan membuat prompt, ketergantungan pada hasil AI, dan berkurangnya kreativitas. Pembahasan menegaskan bahwa efektivitas penggunaan ChatGPT dalam penulisan akademik sangat bergantung pada motivasi belajar siswa, literasi digital, serta peran guru dalam membimbing penggunaan AI yang etis dan reflektif. ARAB يشهد استخدام الذكاء الاصطناعي، وخاصة برنامج شات جي بي تي، في مجال الكتابة الأكاديمية انتشاراً متزايداً في الآونة الأخيرة، مما جعله موضوعاً يثير اهتماماً واسعاً. إذ يساعد شات جي بي تي الطلاب في توليد الأفكار، وتحسين بناء الجمل، وزيادة فعالية عملية الكتابة، إلا أن استخدامه يثير أيضاً عدداً من المخاوف، مثل الاعتماد المفرط، والجوانب الأخلاقية، وتراجع مهارات التفكير النقدي. يهدف هذا البحث إلى استكشاف تصوّرات الطلاب حول استخدام شات جي بي تي في الكتابة الأكاديمية، وكذلك التعرّف على التحديات التي يواجهونها .اعتمدت هذه الدراسة المنهج النوعي بتصميم ظواهري متعالٍ (فينومينولوجيا متعالية). وقد جُمعت البيانات من خلال الملاحظة والمقابلات نصف المهيكلة مع ٢٣ طالباً من الصف الحادي عشر \"الممتاز (A)\" في مدرسة المدرسة العلية ١ نجوج، حيث تم اختيار ستة طلاب منهم كمخبرين رئيسيين. شملت عملية جمع البيانات الملاحظة المباشرة داخل الصف أثناء أنشطة الكتابة، وكذلك المقابلات لفهم تجارب الطلاب ووجهات نظرهم. أظهرت النتائج أن الطلاب يمتلكون تصوّراً إيجابياً تجاه شات جي بي تي بوصفه أداة مفيدة تساعد في توليد الأفكار وتنظيم الكتابة. ومع ذلك، برزت عدة تحديات، منها تحديات تقنية مثل عدم استقرار اتصال الإنترنت، وتحديات معرفية تشمل ضعف مهارة إعداد \"البْرومبت\"، والاعتماد على مخرجات الذكاء الاصطناعي، وتراجع مستوى الإبداع. وتشير المناقشة إلى أن فعالية استخدام شات جي بي تي في الكتابة الأكاديمية تعتمد بدرجة كبيرة على دافع التعلم لدى الطلاب، ومستوى الوعي الرقمي لديهم، ودور المعلم في توجيههم نحو استخدام أخلاقي وواعي للذكاء الاصطناعي.", "source_url": "https:\/\/openalex.org\/W7118152733", "pdf_url": "", "doi": "", "handle": "", "repository_name": "eTheses of Maulana Malik Ibrahim State Islamic University (Maulana Malik Ibrahim State Islamic University)", "source_record_id": "W7118152733", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1212, "title": "Analisis keakuratan penafsiran oleh Chatgpt-5: Studi komparatif penafsiran AI dan tafsir Mafatihul Ghaib", "title_ar": "", "title_en": "", "author": "Isnaini, Inayatul", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Maulana Malik Ibrahim State Islamic University", "university_source": "repository_host", "college": "", "department": "", "country": "Indonesia", "governorate": "", "publication_year": 2025, "language": "id", "research_domain": "nlp", "research_subdomain": "llm", "categories": "deep-learning; nlp; transformers; llm", "keywords": "", "abstract": "ABSTRAK Perkembangan pesat teknologi kecerdasan buatan (Artificial Intelligence\/AI), khususnya ChatGPT yang kini mulai dimanfaatkan dalam bidang studi keislaman, termasuk dalam penafsiran Al-Qur’an. Meskipun AI memiliki kemampuan untuk menyajikan informasi secara cepat dan sistematis, keakuratan hasil tafsir yang dihasilkannya masih perlu diuji secara ilmiah. Oleh karena itu, penelitian ini bertujuan untuk menelaah tingkat akurasi penafsiran yang dihasilkan oleh ChatGPT-5 terhadap ayat Al-Qur’an, dengan membandingkannya terhadap tafsir Mafatih al-Ghaib karya Fakhruddin ar-Razi. Penelitian ini merupakan penelitian kualitatif dengan jenis library research (penelitian kepustakaan). Pendekatan yang digunakan ialah pendekatan tafsir muqāran (komparatif) yang dikombinasikan dengan teori validitas tafsir, guna menilai kesesuaian hasil tafsir ChatGPT-5 dengan prinsip-prinsip penafsiran yang berlaku dalam tradisi tafsir klasik. Data primer berasal dari hasil penafsiran ChatGPT-5 dan kitab tafsir Mafatihul Ghaib, sementara data sekunder meliputi buku, jurnal, dan penelitian terdahulu yang relevan. Hasil penelitian menunjukkan bahwa dari segi keakuratan makna, penafsiran yang dihasilkan oleh ChatGPT-5 dapat dikatakan cukup akurat karena kandungan maknanya sejalan dengan penjelasan yang terdapat dalam Mafatihul Ghaib. Namun, dari aspek sumber rujukan, penafsiran tersebut menunjukkan perbedaan antara kutipan sumber yang disebutkan dengan letak aslinya dalam kitab tafsir klasik, serta ketidaksesuaian data yang menunjukkan bahwa ChatGPT-5 belum sepenuhnya berpijak pada dasar-dasar pemahaman nash yang benar sebagaimana disyaratkan dalam validitas tafsir menurut As-Shabuni. مستخلص البحث بدأ استخدام التطور السريع لتكنولوجيا الذكاء الاصطناعي، وخاصةً ChatGPT، في الدراسات الإسلامية، بما في ذلك تفسير القرآن الكريم. ورغم قدرة الذكاء الاصطناعي على عرض المعلومات بسرعة ومنهجية، إلا أن دقة تفسيراته لا تزال بحاجة إلى اختبار علمي. لذلك، تهدف هذه الدراسة إلى فحص دقة تفسيرات ChatGPT-5 للآيات القرآنية بمقارنتها بتفسير فخر الدين الرازي لكتاب مفاتيح الغيب. هذه الدراسة دراسة نوعية تعتمد على البحث المكتبي. المنهج المستخدم هو منهج تفسير مقارن، مقرون بنظرية صحة التفسير، لتقييم مدى ملاءمة تفسيرات ChatGPT-5 لمبادئ التفسير المعمول بها في تراث التفسير التقليدي. البيانات الأولية مستمدة من تفسير ChatGPT-5 وتفسير مفاتيح الغيب، بينما تشمل البيانات الثانوية الكتب والمجلات والأبحاث السابقة ذات الصلة. تشير نتائج الدراسة إلى أن تفسير ChatGPT-5 دقيقٌ تمامًا من حيث دقة المعنى، نظرًا لتوافقه مع شرح كتاب مفاتيح الغيب. ومع ذلك، من حيث المصادر المرجعية، يُظهر التفسير اختلافات بين الاقتباسات المرجعية وموقعها الأصلي في كتب التفسير القديمة، بالإضافة إلى تباين البيانات، مما يشير إلى أن ChatGPT-5 لا يستند تمامًا إلى أسس الفهم الصحيح للنص، كما هو مطلوب في صحة التفسير عند الشبوني ABSTRACT The rapid development of Artificial Intelligence (AI) technology, particularly ChatGPT, which is now being utilized in the field of Islamic studies, including in the interpretation of the Qur'an. Although AI has the ability to present information quickly and systematically, the accuracy of its interpretive results still needs to be tested scientifically. Therefore, this study aims to examine the level of accuracy of interpretations produced by ChatGPT-5 regarding Qur'anic verses, by comparing them with Mafatih al-Ghaib by Fakhruddin ar-Razi. This research is a qualitative study employing library research methodology. The approach used is the comparative exegesis (tafsir muqāran) method combined with the theory of exegetical validity, in order to assess the conformity of ChatGPT-5's interpretive results with the principles of interpretation prevailing in classical exegetical traditions. Primary data is derived from ChatGPT-5's interpretations and tafsir Mafatihul Ghaib, while secondary data includes books, journals, and relevant previous studies. The research findings indicate that in terms of semantic accuracy, the interpretations produced by ChatGPT-5 can be considered fairly accurate because their semantic content aligns with the explanations found in Mafatih al-Ghaib. The results of the study indicate that in terms of accuracy of meaning, the interpretation produced by ChatGPT-5 can be said to be quite accurate because its meaning is in line with the explanations contained in Tafsir Mafatihul Ghaib. However, from the aspect of reference sources, the interpretation shows differences between the cited source quotes and their original location in classical tafsir books, as well as data discrepancies that indicate that ChatGPT-5 is not fully based on the basics of correct understanding of the text as required in the validity of tafsir according to As-Shabuni.", "source_url": "https:\/\/openalex.org\/W7114933950", "pdf_url": "", "doi": "", "handle": "", "repository_name": "eTheses of Maulana Malik Ibrahim State Islamic University (Maulana Malik Ibrahim State Islamic University)", "source_record_id": "W7114933950", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1216, "title": "Analisis dampak penggunaan ChatGPT dalam kebiasaan pencarian informasi mahasiswa Universitas Brawijaya", "title_ar": "", "title_en": "", "author": "Uswatun Hasanah", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Maulana Malik Ibrahim State Islamic University", "university_source": "repository_host", "college": "", "department": "", "country": "Indonesia", "governorate": "", "publication_year": 2025, "language": "id", "research_domain": "nlp", "research_subdomain": "llm", "categories": "deep-learning; nlp; ai-applications; transformers; llm; arabic-nlp; ai-education", "keywords": "", "abstract": "INDONESSIA Perkembangan teknologi kecerdasan buatan (AI) seperti ChatGPT telah mempengaruhi kebiasaan mahasiswa dalam mencari informasi akademik, sehingga semakin sering menggunakan ChatGPT dalam pemenuhan kebutuhan informasi dan akademik mereka. Penggunaan ChatGPT yang terlalu sering berpotensi dalam mengubah kebiasaan pencarian informasi, pola penggunaan perpustakaan, preferensi terhadap sumber informasi, serta tingkat ketelitian mahasiswa dalam memilah informasi. Penelitian ini bertujuan menganalisis dampak penggunaan ChatGPT terhadap kebiasaan pencarian informasi mahasiswa Universitas Brawijaya. Metode penelitian menggunakan pendekatan kuantitatif dengan penyebaran kuesioner kepada 106 responden. Penelitian ini beracuan pada penelitian lain yang telah dilakukan oleh Karunaratne & Adesina (2023) dengan topik serupa menggunakan lima indikator, yaitu awareness, use and effectiveness, challenges, trust in the function, dan user competence. Hasil penelitian menunjukkan bahwa mahasiswa telah merasakan dampak dari penggunaan ChatGPT dalam proses pencarian informasi. Mahasiswa lebih sering menggunakan ChatGPT sebagai titik awal pencarian informasi karena kemudahan akses dan kecepatan respons, kemudian memverifikasi hasil yang diperoleh melalui sumber ilmiah lain. Sikap kritis ini membantu mahasiswa menghindari kesalahan akibat keterbatasan akurasi ChatGPT sekaligus mencegah ketergantungan berlebihan. Selain itu, pada indikator awareness diketahui bahwa mahasiswa Universitas Brawijaya menyadari manfaat ChatGPT sebagai sarana pencarian informasi cepat, namun pada saat yang sama juga memahami keterbatasannya sehingga menggunakannya secara bijak dan selektif. Pada indikator user competence menunjukkan bahwa penggunaan ChatGPT telah mendorong mahasiswa terbiasa merumuskan pertanyaan lebih spesifik dan meningkatkan kemampuan mengevaluasi kredibilitas informasi. ChatGPT berperan sebagai alat bantu yang mempercepat pencarian informasi, tetapi tetap ditempatkan secara proporsional sebagai pendukung proses belajar, bukan sumber rujukan utama. ENGLISH The development of artificial intelligence (AI) technology such as ChatGPT has influenced students' habits in searching for academic information, so that they increasingly use ChatGPT to fulfill their information and academic needs. Excessive use of ChatGPT has the potential to change information search habits, library usage patterns, preferences for information sources, and the level of accuracy of students in sorting information. This study aims to analyze the impact of ChatGPT usage on the information search habits of students at Brawijaya University. The research method uses a quantitative approach with a questionnaire distributed to 106 respondents. This study refers to another study conducted by Karunaratne & Adesina (2023) on a similar topic using five indicators, namely awareness, use and effectiveness, challenges, trust in the function, and user competence. The results show that students have felt the impact of using ChatGPT in the information search process. Students more often use ChatGPT as a starting point for information search because of its ease of access and speed of response, then verify the results obtained through other scientific sources. This critical attitude helps students avoid mistakes due to ChatGPT's accuracy limitations while preventing excessive dependence. Additionally, the awareness indicator shows that Brawijaya University students recognize the benefits of ChatGPT as a means of quick information retrieval, but at the same time understand its limitations, so they use it wisely and selectively. The user competence indicator shows that the use of ChatGPT has encouraged students to formulate more specific questions and improve their ability to evaluate the credibility of information. ChatGPT serves as a tool that accelerates information retrieval, but it is still placed proportionally as a support for the learning process, not the primary reference source. ARABIC أثر تطور تكنولوجيا الذكاء الاصطناعي مثل شات جي بي تي على عادات الطلاب في البحث عن المعلومات الأكاديمية. يتزايد استخدام الطلاب لـ شات جي بي تي لتلبية احتياجاتهم المعلوماتية والأكاديمية. الاستخدام المتكرر لـ شات جي بي تي من شأنه أن يغير عادات البحث عن المعلومات، وأنماط استخدام المكتبات، وتفضيلات مصادر المعلومات، ومستوى دقة الطلاب في فرز المعلومات. تهدف هذه الدراسة إلى تحليل تأثير استخدام شات جي بي تي على عادات البحث عن المعلومات لدى الطلاب في جامعة براويجايا. تستخدم طريقة البحث نهجًا كميًا مع استبيان تم توزيعه على ١٠٦ مستجيبًا. تشير هذه الدراسة إلى دراسة أخرى أجراها كاروناراتنه وأديسينا (٢٠٢٣) تناقش موضوعًا مشابهًا باستخدام خمسة مؤشرات تم تحليلها، وهي الوعي والاستخدام والفعالية والتحديات والثقة في الوظيفة وكفاءة المستخدم. تظهر النتائج أن الطلاب قد شعروا بالتأثير الحقيقي لاستخدام شات جي بي تي في عملية البحث عن المعلومات.", "source_url": "https:\/\/openalex.org\/W7125093218", "pdf_url": "", "doi": "", "handle": "", "repository_name": "eTheses of Maulana Malik Ibrahim State Islamic University (Maulana Malik Ibrahim State Islamic University)", "source_record_id": "W7125093218", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1222, "title": "Prediction of future Solar and wind Energy Generation through Artificial Neural Network models", "title_ar": "Prediction of future Solar and wind Energy Generation through Artificial Neural Network models", "title_en": "", "author": "احمد علي محمد السبع", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2020, "language": "ar", "research_domain": "deep-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; deep-learning; ai-applications; predictive-analytics; ai-smart-cities", "keywords": "", "abstract": "تقدم هذه الرسائل مراجعة للطرق الحديثة للتنبؤ بالجيل المستقبلي من الطاقة الشمسية وطاقة الرياح من خلال نماذج الشبكات العصبية الاصطناعية في الظروف البيئية المتقلبة, تتمثل الفكرة الرئيسية في استخدام البيانات التاريخية من مزارع الخلايا الشمسية و تور بينات الرياح لتدريب النماذج باستخدام بيانات التدريب التي يتم إجراؤها لفترات مختلفة. وهذا يتطلب بناء نماذج باستخدام الذكاء الاصطناعي (AI) و كذلك باستخدام شبكة عصبية اصطناعية (ANN) و لتحقيق ذلك قمنا ببناء نموذجين هي Feedforward Multi-Layer ANN و Elman Feedback Multi-Layer ANN. استخدم هذا البحث الذكاء الاصطناعي (AI) لقدرته على التعامل مع المعادلات الرياضية المعقدة في الظروف القاسية, نظرا لموثوقيتها ، ودقتها العالية، وإمكانية التنبؤ بها من خلال التدريب على البيانات التاريخية، يتم توقع أربعة نماذج لفترات مختلفة لتوليد الطاقة على مرحلتين. المرحلة الأولى: عملية التدريب على البيانات التاريخية لمدة شهر والتنبؤ بها لمدة يوم واحد في المستقبل يطبق على النماذج الأربعة السابقة وإيجاد أفضل نموذج لشبكة ANN للتنبؤ بكمية توليد الطاقة بحيث يتم أخذ النموذج الذي يتنبأ بكمية توليد الطاقة بدقة عالية و أفضل أداء وأقل معدل خطأ من بين النموذجين ثم ننتقل الى المرحلة الثانية. المرحلة الثانية: إنها مرحلة تدريب على البيانات لمدة شهر و التنبؤ لمدة سبع أيام بشكل تدريجي لإيجاد أفضل أداء و لتقليل الحد الأدنى من الأخطاء المرتبطة بالتنبؤ بحمل الطاقة الكهربائية من أجل تقديم تقرير دقيق للغاية للتنبؤ بكمية توليد الطاقة في المستقبل في تاريخ محدد ليوم محدد بعد أسبوع أو شهر أو سنة، من أجل الموثوقية والتكامل الفعال للشبكة لإيجاد حلول لمعالجة أزمة الطاقة, وبالتالي خفض التكلفة المالية في المنطقة بسبب استيراد النفط والغاز لإنتاج حمولة الكهرباء. علاوة على ذلك، للحد من التلوث البيئي للحد من النفايات من استخراج النفط والغاز والفحم, هناك فائدة أخرى هي وجود الخلايا الشمسية لخدمة بعض المناطق بالطاقة الكهروضوئية, هذه إحدى الفوائد عندما نعرف استهلاك المنطقة بدقة. توفر الشركة الكهربائية هذه المنطقة بشكل أفضل إما عن طريق زيادة الخلايا الشمسية و تور بينات الرياح أو زيادة واردات الغاز وتجنب فصل الكهرباء وتوفير الصفقات والحفاظ على كمية الاستيراد للغاز لتوليد الطاقة, تمكن النموذجين المقترحين لهذا العمل من تقديم نتائج أفضل من الأعمال السابق (Idris Khana, Honglu Zhua, ET al. 2017, Luis Carlos Parra Raffan , Andres Romero, ET al. 2018 ). بدقة تصل إلى 96.6?", "source_url": "https:\/\/openalex.org\/W3212747612", "pdf_url": "", "doi": "", "handle": "", "repository_name": "", "source_record_id": "W3212747612", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1223, "title": "Implementasi Particle Swarm Optimization pada NPC Hunter untuk pencarian NPC Hewan Hunter of Forest", "title_ar": "Implementasi Particle Swarm Optimization pada NPC Hunter untuk pencarian NPC Hewan Hunter of Forest", "title_en": "", "author": "Achmad Ihwany", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Maulana Malik Ibrahim State Islamic University", "university_source": "repository_host", "college": "", "department": "", "country": "Indonesia", "governorate": "", "publication_year": 2018, "language": "ar", "research_domain": "machine-learning", "research_subdomain": "optimization", "categories": "machine-learning; optimization", "keywords": "", "abstract": "مستخلص البحث تحسين سرب الجسيمات (PSO) ، والذي يحاكي شخصية مجموعة من الطيور أو الأسماك. تحاكي خوارزمية PSO السلوك الاجتماعي لهذا الكائن الحي. يتكون السلوك الاجتماعي من العمل الفردي وتأثير الأفراد الآخرين في المجموعة. كلمة الجسيمات تثبت ، على سبيل المثال ، طائرًا في قطيع من الطيور. في كل شخصية فردية أو جسيمية باستخدام ذكائه الخاص وأيضاً بسبب سلوكه الجماعي المشترك. لذلك ، إذا كان أحد الجسيمات أو طائر واحد يحصل على أفضل مسار أو أسرع مسار إلى مصدر الغذاء ، فسيجد الطائر الآخر أو المجموعة الأخرى أيضًا المسار الأسرع أو الأسرع ، على الرغم من أن موقعه بعيد جدًا في المجموعة قبل ، أحرف غير المشغِّل (NPC) أو تسمى أيضًا بالكيانات هي كيان في اللعبة لا يتحكم فيه المشغل مباشرة. يتم التحكم في NPC تلقائيًا بواسطة الكمبيوتر. لتكون قادرة على الحصول على سلوك ذكي من المجلس الوطني للمرأة استخدام الذكاء الاصطناعي أو الذكاء الاصطناعي (منظمة العفو الدولية). يتم استخدام. AIS في NPC من خلال توفير خوارزميات محددة وفقًا للسلوك الذكي المتوقع تتناول هذه الدراسة تطبيق خوارزمية تحسين سرع الجسيمات لتحديد سلوك صائد NPC في لعبة \" صياد الغابة. \" حتى تتمكن من تحديد سلوك الصياد في لعبة FPS صياد الغابة المعتمدة على سطح المكتب نتيجة هذا البحث ، هذا البحث ينجح أقصر طريق إلى لعبة الصيد المطبق على NPC Hunter بشكل حيوي باستخدام خوارزمية سرب تحسين الجسيمات ABSTRACT Particle swarm optimization (PSO), which mimics the character of a group of birds or fish. The PSO algorithm mimics the social behavior of this organism. Social behavior consists of individual action and the influence of other individuals in a group. The word particle proves, for example, a bird in a flock of birds. In each individual or particle character using his own intelligence and also due to his collective joint behavior. Therefore, if one of the particles or one bird gets the best path or the fastest path to the food source, the other bird or other group will also find the fastest or fastest route, although their location is far in the group before . Non-Player Characters (NPC) or also called agents are an in-game entity that is not directly controlled by the player. NPC is automatically controlled by computer. To be able to obtain intelligent behavior from NPC used artificial intelligence or Artificial Intelligence (AI). The use of AI in NPC is done by providing specific algorithms according to the expected intelligent behavior. This study discusses the Implementation of Particle Swarm Optimization Algorithm for Determining the Behavior of NPC Hunter in \"Hunter Of Forest\" Game In order to be able to determine the hunter behavior in the desktop-based FPS Hunter Of Forest game. The result of this research is, This research succeed the shortest path to game hunting applied to NPC Hunter dynamically using Particle Swarm Optimization Algorithm ABSTRAK Particle swarm optimization (PSO), yang meniru karakter sekelompok burung atau ikan. Algoritma PSO meniru perilaku sosial organisme ini. Perilaku sosial terdiri dari aksi individu dan pengaruh dari individu-individu lain dalam suatu kelompok. Kata partikel membuktikan, misalnya, seekor burung dalam kawanan burung. Pada masing-masing individu atau partikel berkarakter memakai kecerdasannya (intelligence) sendiri dan juga diakibatkan tingkah laku gabungan kolektifnya. Maka dari itu, apabila salah satu dari partikel atau satu burung mendapatkan jalan yang terbaik atau jalan yang tercepat untuk menuju ke sumber makanan, maka burung yang lain atau kelompok yang lain juga akan menemukan jalur tebaik atau tercepat tadi, meskipun lokasinya mereka jauh di kelompok tadi. Non-Player Characters (NPC) atau disebut juga agen adalah suatu entitas dalam game yang tidak dikendalikan secara langsung oleh pemain. NPC dikendalikan secara otomatis oleh komputer. Untuk dapat memperoleh perilaku cerdas dari NPC digunakan kecerdasan buatan atau Artificial Intelligence (AI). Penggunaan AI pada NPC dilakukan dengan pemberian algoritma khusus sesuai dengan perilaku cerdas yang diharapkan. Penelitian ini membahas mengenai Implementasi Algoritma Particle Swarm Optimization Untuk Menentukan Perilaku NPC Hunter Pada Game “Hunter Of Forest” Dengan tujuan agar dapat menentukan prilaku hunter pada game FPS Hunter Of Forest yang berbasis dekstop. Hasil dari penelitian ini adalah, Penelitian ini berhasil jalur terpendek menuju hewan buruan yang diterapkan pada NPC Hunter secara dinamis menggunakan Algoritma Particle Swarm Optimization.", "source_url": "https:\/\/openalex.org\/W2907762393", "pdf_url": "http:\/\/etheses.uin-malang.ac.id\/12535\/1\/11650087.pdf", "doi": "", "handle": "", "repository_name": "eTheses of Maulana Malik Ibrahim State Islamic University (Maulana Malik Ibrahim State Islamic University)", "source_record_id": "W2907762393", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1229, "title": "Reinforcement Learning-Based Ordering", "title_ar": "", "title_en": "Reinforcement Learning-Based Ordering", "author": "Samuel Dahn ; Theodor Hallbeck", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Lund University", "university_source": "repository_host", "college": "", "department": "", "country": "Sweden", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "reinforcement-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; reinforcement-learning; predictive-analytics", "keywords": "", "abstract": "This thesis investigates the applicability and performance of reinforcement learn- ing (RL) for inventory management of spare parts in a single-echelon setting. Traditional inventory control methods, such as (R,Q)-policies, typically decou- ple forecasting and ordering decisions, which may lead to suboptimal system- wide performance, particularly in environments characterized by intermittent and stochastic demand. To address these limitations, a reinforcement learning-based ordering model is developed and evaluated using both simulated and real-world data provided by Syncron AB. The proposed model is a Dueling Double Deep Q-Network (DDDQN) architecture incorporating enhancements such as prioritized expe- rience replay, experience buffer modification, and early stopping to improve training efficiency and stability. The performance of the RL-model is benchmarked against a standard peri- odic review (R,Q)-policy using backorder and holding cost. Results indicate that while RL demonstrates potential in capturing complex system dynamics and achieving competitive performance in certain scenarios, its effectiveness is highly dependent on data availability, demand characteristics, and computa- tional constraints. In particular, challenges related to intermittent demand, policy convergence, and generalization to unseen environments restrict consis- tent outperformance of traditional methods. The findings suggest that RL based approaches can complement existing inven- tory control methods but cannot yet fully replace them in practical applications. This highlights the importance of hybrid decision frameworks that leverage the strengths of both data-driven and analytical methods. The thesis concludes with recommendations for future development, including model scalability, improved data utilization, and more robust evaluation across varying demand patterns.", "source_url": "https:\/\/openalex.org\/W7165431631", "pdf_url": "https:\/\/lup.lub.lu.se\/student-papers\/record\/9238531\/file\/9238536.pdf", "doi": "", "handle": "", "repository_name": "Lund University Publications Student Papers (Lund University)", "source_record_id": "W7165431631", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1231, "title": "Learning to Plan with Reinforcement Learning", "title_ar": "", "title_en": "Learning to Plan with Reinforcement Learning", "author": "Ming Hay Chung", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Cambridge", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "nlp; reinforcement-learning; responsible-ai; llm; xai", "keywords": "", "abstract": "This thesis investigates the domain of learning to plan within reinforcement learning. Traditional planning algorithms, such as Monte Carlo Tree Search, are typically handcrafted and lack the flexibility inherent in biological planning systems. To address this limitation, the Thinker algorithm is proposed. A central theme of this work is the shift in focus from designing specialized planning algorithms to augmenting the environment. By treating planning as an explicit action within an augmented environment, the agent is enabled to learn to plan using standard model-free reinforcement learning. Additionally, this thesis examines the interpretability of planning behaviour of RL agents trained by different algorithms. Finally, this thesis extends the Thinker algorithm to Large Language Models trained on mathematical reasoning tasks, demonstrating that similarly structured augmented environments significantly reduce computational costs while maintaining competitive performance on reasoning tasks.", "source_url": "https:\/\/openalex.org\/W7159592709", "pdf_url": "https:\/\/www.repository.cam.ac.uk\/bitstreams\/83ddc738-d30a-4859-a414-c09e906b7018\/download", "doi": "10.17863\/cam.129818", "handle": "", "repository_name": "Apollo (University of Cambridge)", "source_record_id": "W7159592709", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1235, "title": "Explaining reinforcement learning", "title_ar": "", "title_en": "", "author": "Léo Saulières", "supervisor": "", "degree_type": "phd", "degree_evidence": "theses.fr: الفهرس الوطني الفرنسي لأطاريح الدكتوراه", "university": "", "university_source": "", "college": "", "department": "", "country": "France", "governorate": "", "publication_year": 2024, "language": "fr", "research_domain": "computer-vision", "research_subdomain": "xai", "categories": "computer-vision; reinforcement-learning; robotics; responsible-ai; ai-applications; xai; ai-agriculture", "keywords": "", "abstract": "Explication de l'apprentissage par renforcement S dernières années, les modèles issus de l'Intelligence Artificielle (IA) ont connu une progression impressionnante tant sur la précision de leurs résultats que sur l'amplitude de leurs applications. Cette progression s'explique en partie par l'utilisation de réseaux de neurones permettant de résoudre efficacement diverses tâches en se basant sur un ensemble de données. Les différentes avancées en IA prédictive (par opposition à l'IA analytique qui s'intéresse à la représentation des connaissances et à la formalisation du raisonnement) ont été mises au service de domaines variés comme l'agriculture, la médecine ou encore l'éducation. On remarque également que l'IA prédictive est de plus en plus présente dans notre vie de tous les jours, que cela soit pour de la recommandation de musique avec Spotify, de la génération d'image avec Midjourney ou encore de l'aide à la synthèse de réunion avec Leexi. Cette omniprésence entraîne un questionnement vis-à-vis de la régulation de ces modèles. Avec, entre autres, l'utilisation des réseaux de neurones, les modèles performants s'apparentent à des boîtes noires. L'emploi de ces modèles opaques pour des domaines à risque, tels que la médecine ou les véhicules autonomes, requiert un contrôle spécifique avant la mise à disposition aux utilisateurs. Ainsi, des institutions proposent de réglementer l'IA (par exemple, la Commission Européenne avec l'AI Act). Ces réglementations mettent en avant l'obligation de vérifier, expliquer et analyser les modèles opaques, particulièrement pour les applications à haut risque. De plus, l'utilisateur doit avoir le droit de demander des explications sur la prise de décision d'un modèle, comme le rejet d'attribution d'un prêt par une banque. Ce besoin de transparence a entraîné le développement de l'IA explicable (XAI), à savoir le domaine proposant des méthodes pour rendre les modèles d'IA moins opaque, voire transparent. Nous nous focalisons dans cette thèse sur l'explication d'un type d'IA appelé Apprentissage par Renforcement (RL), qui consiste à apprendre un comportement adéquat dans un environnement au travers d'essais-erreurs. L'agent (ou modèle d'IA) apprend à effectuer, depuis chaque état de l'environnement, une action optimale pour la réalisation d'une tâche spécifique. Cette thèse s'inscrit dans le domaine de l'Apprentissage par Renforcement explicable (XRL), un domaine en pleine expansion qui regroupe les méthodes d'explication du RL. Nos deux contributions majeures concernent la proposition d'une taxonomie originale de l'état de l'art du XRL et la création et l'analyse de 4 méthodes pour le XRL. Notre taxonomie catégorise les différentes méthodes du XRL par rapport à la cible de l'explication et à la manière de l'expliquer. Un nouvel environnement multi-agent de RL, intitulé 'Drone Coverage', a été implémenté pour tester nos méthodes. Celles-ci relèvent de deux catégories différentes selon notre taxonomie. Avec 'Scenario eXplanation' (SXp) et 'Expected States eXplanation' (ESX), nous expliquons l'action d'un agent effectuée depuis un état donné en fournissant des conséquences attendues. SXp résume le futur en 3 séquences d'état-action appelées scénarios. ESX compacte les états que l'agent peut atteindre à un certain horizon en un sous-objectif concis et représentatif. Avec 'History eXplanation based on Predicates' (HXP) et 'Backward-HXP' (B-HXP), nous expliquons une séquence d'état-action de l'agent en fournissant un ensemble d'éléments importants de cette séquence. HXP extrait d'une séquence d'état-action courte, les actions les plus importantes pour la réalisation d'un prédicat défini. A partir d'une séquence d'état-action longue, B-HXP fournit un ensemble de prédicats et actions importantes associées en procédant récursivement depuis la fin de la séquence.", "source_url": "https:\/\/openalex.org\/W7152368558", "pdf_url": "https:\/\/theses.hal.science\/tel-05080042\/document", "doi": "10.70675\/ef57c459z3e5bz437czbc31zec8bfff34d87", "handle": "", "repository_name": "theses.fr (ABES)", "source_record_id": "W7152368558", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1236, "title": "Explorations of In-Context Reinforcement Learning", "title_ar": "", "title_en": "Explorations of In-Context Reinforcement Learning", "author": "Ethan Brooks", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Michigan", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; nlp; reinforcement-learning; self-semi-supervised; rnn; transformers; llm", "keywords": "", "abstract": "In-Context Learning describes a form of learning that occurs when a model accumulates information in its context or memory. For example, a Long Short-Term Memory unit (LSTM) can rapidly adapt to a novel task as input\/target exemplars are fed into it. While in-context learning of this kind is not a new discovery, recent work has demonstrated the capacity of large \"foundation\" models to acquire this ability \"for free,\" by training on large quantities of semi-supervised data, without the sophisticated (but often unstable) meta-objectives proposed by many earlier papers #cite(< finn2017model>)#cite(< stadie2018some>)#cite(< rakelly2019efficient>). In this work we explore several algorithms which specialize in-context learning based on semi-supervised methods to the reinforcement learning (RL) setting. In particular, we explore three approaches to in-context learning of value (expected cumulative discounted reward). The first of these methods demonstrates a method for implementing policy iteration, a classic RL algorithm, using a pre-trained large language model (LLM). We use the LLM to generate planning rollouts and extract monte-carlo estimates of value from them. We demonstrate the method on several small, text-based domains and present evidence that the LLM can generalize to unseen states, a key requirement of learning in non-tabular settings. The second method imports many of the ideas of the first, but trains a transformer model directly on offline RL data. We incorporate Algorithm Distillation (AD) #cite(< laskin2022context>), another method for in-context reinforcement learning that directly distills the improvement operator from data that includes behavior ranging from random to optimal. Our method combines the benefits of AD with the policy iteration method proposed in our previous work and demonstrates benefits in performance and generalization. Our third method proposes a new method for estimating value. Like the previous methods, this one implements a form of policy iteration, but eschews monte-carlo rollouts for a new approach to estimating value. We train a network to estimate Bellman updates and iteratively feed its outputs back into itself until the estimate converges. We find that this iterative approach improves the capability of the value estimates to generalize and mitigates some of the instability of other offline methods.", "source_url": "https:\/\/openalex.org\/W4398230562", "pdf_url": "", "doi": "10.7302\/23097", "handle": "", "repository_name": "Deep Blue (University of Michigan)", "source_record_id": "W4398230562", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1243, "title": "Safe Hierarchical Reinforcement Learning: Hierarchical Reinforcement Learning with Safe Space Exploration", "title_ar": "", "title_en": "Safe Hierarchical Reinforcement Learning: Hierarchical Reinforcement Learning with Safe Space Exploration", "author": "Stephen Verbist", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Delft University of Technology", "university_source": "repository_host", "college": "", "department": "", "country": "Netherlands", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "reinforcement-learning", "research_subdomain": "", "categories": "reinforcement-learning", "keywords": "", "abstract": "Reinforcement Learning is a much researched topic for autonomous machine behavior and is often applied to navigation problems. In order to deal with growing environments and larger state\/action spaces, Hierarchical Reinforcement Learning has been introduced. Unfortunately learning from experience, which is central to Reinforcement Learning, makes guaranteeing safety a complex problem. This paper demonstrates an approach, named the actor-judge approach, to make the exploration safer while imposing as few as possible restrictions on the agent. The approach combines ideas from the fields of Hierarchical Reinforcement Learning and Safe Reinforcement Learning to develop a Safe Hierarchical Reinforcement Learning algorithm. The algorithm is tested in a simulated environment where the agent represents an Unmanned Aerial Vehicle able to move laterally in four directions using quadridirectional range sensors to establish a relative position. Although this approach does not guarantee the agent to never explore unsafe areas of the state domain, results show the actor-judge method increases agent safety and can be used on multiple levels an HRL agent hierarchy.", "source_url": "https:\/\/openalex.org\/W2614018942", "pdf_url": "https:\/\/repository.tudelft.nl\/file\/File_10ae1c30-9e76-4af3-982f-34b67806dad2", "doi": "", "handle": "", "repository_name": "Research Repository (Delft University of Technology)", "source_record_id": "W2614018942", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1254, "title": "PERFORMANCE ANALYSIS OF TRANSFORMER-BASED LANGUAGE MODELS FOR DETECTION OF PHISHING EMAILS", "title_ar": "", "title_en": "PERFORMANCE ANALYSIS OF TRANSFORMER-BASED LANGUAGE MODELS FOR DETECTION OF PHISHING EMAILS", "author": "Chidozie Princewill Arannonu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; nlp; ai-applications; transfer-learning; transformers; llm; ai-cybersecurity", "keywords": "", "abstract": "This thesis evaluates the performance and adversarial robustness of four transformer-based language models; BERT, RoBERTa, DistilBERT, and GPT-2 which were fine-tuned for binary phishing email classification. Two preprocessing configurations (Basic and Advanced) and five training data fractions are systematically compared using a temporally partitioned dataset that separates pre-2015 training data from 2015–2024 deployment data, simulating real-world generalization across evolving phishing tactics. DistilBERT with Advanced preprocessing achieves the best overall performance, reaching 97.26% accuracy and a weighted F1-score of 0.972. The study further conducts adversarial robustness evaluation using four attack strategies from the TextAttack framework namely; DeepWordBug, Pruthi, TextBugger, and TextFooler which found out that character-level attacks achieve success rates of up to 22.44%. Finally, applying Advanced preprocessing to adversarially perturbed inputs recovers post-attack accuracy to near-clean-data levels across all models, establishing preprocessing as a deployable defense mechanism rather than merely a data cleaning step.", "source_url": "https:\/\/openalex.org\/W7160619207", "pdf_url": "", "doi": "10.25394\/pgs.32204841", "handle": "", "repository_name": "Purdue", "source_record_id": "W7160619207", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1256, "title": "SISTEM PENILAIAN FORUM DISKUSI MAHASISWA PADA E�LEARNING BERBASIS MODEL BERT", "title_ar": "", "title_en": "", "author": "Ahmad Habibi ; Endina Putri Purwandari ; Widhia oktoeberza", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Bengkulu", "university_source": "repository_host", "college": "", "department": "", "country": "Indonesia", "governorate": "", "publication_year": 2025, "language": "id", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; nlp; ai-applications; transfer-learning; transformers; ai-education", "keywords": "", "abstract": "Penelitian ini bertujuan untuk merancang dan mengembangkan sistem penilaian otomatis forum diskusi mahasiswa pada platform E-Learning dengan memanfaatkan model BERT (Bidirectional Encoder Representations from Transformers). Penelitian ini dilatarbelakangi oleh tantangan dalam menilai partisipasi mahasiswa dalam forum diskusi E-Learning, nyatanya dengan metode tradisional sering kali bersifat subjektif dan memakan waktu. Sedangkan model BERT dikenal dengan kemampuannya dalam memahami konteks teks secara mendalam, sehingga diharapkan dapat meningkatkan objektivitas dan efisiensi dalam proses penilaian. Penelitian ini mencakup beberapa tahapan, mulai dari analisis kebutuhan, perancangan sistem, pengembangan fine-tuning model BERT, hingga evaluasi dengan membandingkan hasil penilaian otomatis dan manual. Sistem ini dirancang untuk menilai kontribusi mahasiswa berdasarkan kompleksitas dan relevansi kontennya, dengan tujuan meningkatkan interaksi serta kualitas diskusi dalam E-Learning. Hasil yang diharapkan adalah sistem penilaian yang lebih akurat, efisien, dan mampu memberikan umpan balik yang bermanfaat bagi mahasiswa maupun dosen. Selain itu, penelitian ini diharapkan dapat memberikan kontribusi signifikan dalam pengembangan teknologi penilaian otomatis berbasis Natural Language Processing (NLP) di bidang pendidikan. Kata kunci: E-Learning, BERT, Penilaian Otomatis, Forum Diskusi, Teknologi Pendidikan.", "source_url": "https:\/\/openalex.org\/W7126128896", "pdf_url": "", "doi": "", "handle": "", "repository_name": "University of Bengkulu Scholar Repository (University of Bengkulu)", "source_record_id": "W7126128896", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1261, "title": "Improving Code Quality Using Fine-tuning Large Language Models", "title_ar": "", "title_en": "Improving Code Quality Using Fine-tuning Large Language Models", "author": "Quang Hieu Nguyen", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Tampere University", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; deep-learning; nlp; generative-ai; reinforcement-learning; transfer-learning; transformers; llm", "keywords": "", "abstract": "Large language models has demonstrated significant capabilities in solving real-life problems by the means of generating human-like responses to input text in the similar format. However, these generic languages exhibit weakness when the context is not fully provided, resulting a need for model customization to enhance their performance in specialized fields. The fine-tuning is implemented using techniques inspired by machine learning methods. The two well-known fine-tuning methods are Supervised Fine-Tuning (STF), based on the concept of supervised learning, and Reinforcement Learning from Human Feedback (RLHF) introduced by OpenAI. The resource-intensive nature of RLHF makes it challenging to replicate the results achieved by Chat Generative Pre-trained Transformer (ChatGPT) models. Therefore, a new technique named direct preference optimization (DPO) has been introduced lately, incorporating similar concepts to RLHF while reducing required time and labor. Given several studies have not addressed the improvement of DPO compared to other fine-tuning methods and its interplay with them, this thesis employs STF, DPO and their combination to fine-tune the base LLAMA 3.1 model, whose size is 8 billion parameters, in order to improve code quality generated by the LLMs. These fine-tuned models are then evaluated using Bi-Lingual Evaluation Understudy (BLEU) and Bidirectional Encoder Representations from Transformer (BERT) metrics, along with their code-specific derivatives. Despite the outcomes of the metrics cannot determine the most effective fine-tuning technique, they clearly highlight the significant influence of dataset utilized during the fine-tuning process. The repository containing the code and data for the research can be accessed via GitHub platform.", "source_url": "https:\/\/openalex.org\/W7133076851", "pdf_url": "https:\/\/trepo.tuni.fi\/bitstream\/10024\/161629\/2\/NguyenDuc.pdf", "doi": "", "handle": "", "repository_name": "Tampere University Institutional Repository (Tampere University)", "source_record_id": "W7133076851", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1264, "title": "Microbial named entity recognition using BERT models", "title_ar": "", "title_en": "Microbial named entity recognition using BERT models", "author": "Brian K Rao", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Illinois at Urbana-Champaign", "university_source": "repository_host", "college": "", "department": "", "country": "United States", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; nlp; medical-ai; ai-applications; transfer-learning; data-mining; transformers; ai-industry", "keywords": "", "abstract": "Bacteria are critical subjects of microbiological research that span many rapidly-growing fields of study.The development and widespread application of high-throughput sequencing has led to more microbial data being collected in recent years than ever before.This study investigates the capabilities of the popular Natural Language Processing (NLP) model Bidirectional Encoder Representations from Transformers (BERT) on the relatively understudied text mining domain of microbiology.This is done by fine-tuning a variety of BERT models (BERT, DistilBERT, SciBERT, BioBERT, PubMedBERT) on the Bacteria Biotope 2019 Open Shared Task (BB2019-OST) corpus of annotated microbial research text and evaluating the best performing models on the Named Entity Recognition (NER) task.Following this, an in-depth error analysis was conducted to gain insights into BERT's entity recognition capabilities.Finally, to investigate performance capabilities further, learning rate and batch size hyperparameters were tuned to increase F1-score.The best BERT model in the comparison was BioBERT, earning an F1-score of 73.82 (±1.04) with default hyperparameters, and 75.35 (±0.62) with tuned hyperparameters.BioBERT had better F1-scores and entity-level statistics, despite PubMedBERT ranking high in biomedical NLP benchmarks.This suggests that the generality of the pretraining corpora of BERT models is particularly important for text mining in the microbial domain.", "source_url": "https:\/\/openalex.org\/W7220763410", "pdf_url": "https:\/\/www.ideals.illinois.edu\/items\/125125\/bitstreams\/410938\/data.pdf", "doi": "", "handle": "", "repository_name": "Illinois Digital Environment for Access to Learning and Scholarship (University of Illinois at Urbana-Champaign)", "source_record_id": "W7220763410", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1271, "title": "Transformer-based kernels for Natural Language Processing", "title_ar": "", "title_en": "Transformer-based kernels for Natural Language Processing", "author": "David Morán Pomés", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; nlp; transfer-learning; rnn; transformers", "keywords": "", "abstract": "Kernel methods (KM) and Artificial Neural Networks (ANN) are two of the main families of methods that are more widely spread nowadays in the world of Machine Learning, being the Support Vector Machine (SVM) and the Deep Learning Neural Network (DLNN) its most known members. One of the main benefits of both families is how they can be adapted to work with different kinds of input data, even though they do it in a different way: KM rely on different kernel functions and ANN rely on specific network architectures. In the specific world of Natural Language Processing, both families have specific resources to deal with texts as inputs. While KM rely on specific kernels like the string kernel, ANN rely on Recurrent Neural Network (RNN) architectures and, more recently, on Transformer architectures. Even though KM were the first ones to dive into the NLP world, ANN are nowadays the state-of-the-art in all kinds of NLP tasks. The main reason of this success is the fact that ANN perform as a nonlinear parametric feature extractor, which processes the text in a complex way that lets the model learn an adequate representation of the input. On the other side, kernel functions do only rely on fixed similarity measures to contain that information, which is less flexible and powerful. This is the reason why we developed a new kernel function, called Transformer kernel. This kernel function is a parametric model like an ANN, but instead of computing a nonlinear representation of the input, it directly computes a similarity measure that corresponds to an inner product in some implicit feature space. The main idea of this model is that it can be pre-trained in a generic task using the whole Wikipedia corpus (like the different Transformer models for NLP tasks), but the resulting model can be directly used as a kernel function for any kind of kernel method (like the SVM). In order to do so, we build this model by stacking two different components: A Transformer Encoder block (based on the BERT architecture) and a newly designed kernel function called Attention Kernel. By using this kernel function with real text datasets, we have shown that it provides results that are much better than the ones obtained by the use of the string kernel. We have also seen that, due to the fact that our model is built as a Neural Network and can be executed on a GPU, the Transformer kernel can compute the kernel matrix of a dataset much faster than the classical string kernel. Results are very promising, and suggest that further study of the behaviour of this kernel function with different datasets for pre-training and different final tasks will be very interesting. Moreover, it has to be considered if it is possible to adapt this kernel in order to combine the text input with other kinds of data (like numerical or categorical variables), or if the kernel can be fine-tuned into a specific task before it is used as a kernel function.", "source_url": "https:\/\/openalex.org\/W3138190555", "pdf_url": "", "doi": "", "handle": "", "repository_name": "LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas)", "source_record_id": "W3138190555", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1277, "title": "Is deep learning the same as deep learning? : perspectives on the construct of deep learning", "title_ar": "", "title_en": "Is deep learning the same as deep learning? : perspectives on the construct of deep learning", "author": "Karianne Strauman", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Norway", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "deep-learning", "research_subdomain": "", "categories": "deep-learning", "keywords": "", "abstract": "", "source_url": "https:\/\/openalex.org\/W2884223569", "pdf_url": "", "doi": "", "handle": "", "repository_name": "BIBSYS Brage (BIBSYS (Norway))", "source_record_id": "W2884223569", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1288, "title": "Rhythm interpretation using deep learning neural networks", "title_ar": "", "title_en": "Rhythm interpretation using deep learning neural networks", "author": "Daniel Eidsvåg", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Norway", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "medical-ai", "research_subdomain": "ai-industry", "categories": "machine-learning; deep-learning; medical-ai; ai-applications; ai-industry", "keywords": "", "abstract": "Out-of-hospital cardiac arrest (OHCA) is a leading cause of death in the industrialized world, with an estimated annual incidence that varies by 52.5 (in Asia), 86.4 (in Europe), 98.1 (in North America), and 111.9 (in Australia) per 100,000 person-years. Lethal ventricular arrhythmias are the most frequent causes of OHCA. A defibrillation shock is an effective treatment, and early defibrillation is one of the key factors in survival from OHCA. However, chest compressions, ventilations, and drug play a key role in the treatment of cardiac arrest. Under resuscitation, automated external defibrillator (AED) require peri-shock pauses to analyze for a shockable rhythm. This peri-shock are associated with a decrease in survival to hospital discharge. The objective of this thesis is to determine if different artificial neural network (ANN) structures can be used as a classifier, to determine the underlying heart rhythm under chest compressions to remove peri-shock pauses during cardiac arrest. The analysis is conducted from data obtained from 394 OHCA patients, where two datasets were used. Both containing 3-second segments with clinical rhythm annotations resulting in 2446 and 422415 segments. Results in this thesis suggest that there is no clear best method in the different neural network methods for ECG data. However, FNN demonstrates the most promising results with an accuracy of 52.30%. This result emphasizes the problem regarding classification of ECG data with compression artifacts.", "source_url": "https:\/\/openalex.org\/W2756701130", "pdf_url": "", "doi": "", "handle": "", "repository_name": "BIBSYS Brage (BIBSYS (Norway))", "source_record_id": "W2756701130", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1289, "title": "Hand gesture recognition using deep learning neural networks", "title_ar": "", "title_en": "Hand gesture recognition using deep learning neural networks", "author": "Norah M. Alnaim", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Brunel University London", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "machine-learning; deep-learning; computer-vision; nlp; medical-ai; predictive-analytics; cnn; video-analysis; speech-recognition", "keywords": "", "abstract": "Human Computer Interaction (HCI) is a broad field involving different types of interactions including gestures.Gesture recognition concerns non-verbal motions used as a means of communication in HCI.A system may be utilised to identify human gestures to convey information for device control.This represents a significant field within HCI involving device interfaces and users.The aim of gesture recognition is to record gestures that are formed in a certain way and then detected by a device such as a camera.Hand gestures can be used as a form of communication for many different applications.It may be used by people who possess different disabilities, including those with hearing-impairments, speech impairments and stroke patients, to communicate and fulfil their basic needs.Various studies have previously been conducted relating to hand gestures.Some studies proposed different techniques to implement the hand gesture experiments.For image processing there are multiple tools to extract features of images, as well as Artificial Intelligence which has varied classifiers to classify different types of data.2D and 3D hand gestures request an effective algorithm to extract images and classify various mini gestures and movements.This research discusses this issue using different algorithms.To detect 2D or 3D hand gestures, this research proposed image processing tools such as Wavelet Transforms and Empirical Mode Decomposition to extract image features.The Artificial Neural Network (ANN) classifier which used to train and classify data besides Convolutional Neural Networks (CNN).These methods were examined in terms of multiple parameters such as execution time, accuracy, sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood, negative likelihood, receiver operating characteristic, area under ROC curve and root mean square.This research discusses four original contributions in the field of hand gestures.The first contribution is an implementation of two experiments using 2D hand gesture video where ten different gestures are detected in short and long distances using an iPhone 6 Plus with 4K resolution.The experiments are performed using WT and EMD for feature extraction while ANN and CNN for classification.The second contribution comprises 3D hand gesture video experiments where twelve gestures are recorded using holoscopic imaging system camera.The third contribution pertains experimental work carried out to detect seven common hand gestures.Finally, disparity experiments were performed using the left and the right 3D hand gesture videos to discover disparities.The results of comparison show the accuracy results of CNN being 100% compared to other techniques.CNN is clearly the most appropriate method to be used in a hand gesture system.", "source_url": "https:\/\/openalex.org\/W3085860926", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Brunel University Research Archive (BURA) (Brunel University London)", "source_record_id": "W3085860926", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1297, "title": "Medical image analysis with machine learning techniques and deep learning neural networks", "title_ar": "", "title_en": "", "author": "Γεώργιος Γιαννουλόπουλος ; Giannoulopoulos, George", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Piraeus", "university_source": "repository_host", "college": "", "department": "", "country": "Greece", "governorate": "", "publication_year": 2020, "language": "el", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; medical-imaging", "keywords": "", "abstract": "Οι πρόσφατες εξελίξεις στον τομέα της μηχανικής μάθησης στη μηχανική όραση θα εξεταστούν ,ενώ ένα υποσύνολο αυτών των αλγορίθμων θα δοκιμαστεί στο πρόβλημα της ταξινόμησης δερματικών παθήσεων. Πιο συγκεκριμένα, η απόδοση των βαθιών νευρωνικών δικτύων στη ταξινόμηση εικόνων θα παρουσιαστεί και δοκιμαστεί. Τέλος, μια συγκεκριμένη οικογένεια βαθιών νευρωνικών δικτύων που λέγονται EfficientNets θα εκπαιδευτεί στο σύνολο δεδομένων του ISIC challenge του 2019.", "source_url": "https:\/\/openalex.org\/W3081787540", "pdf_url": "", "doi": "10.26267\/unipi_dione\/239", "handle": "", "repository_name": "Dione (University of Piraeus)", "source_record_id": "W3081787540", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1299, "title": "Autonomous control of a mobile robot with incremental deep learning neural networks", "title_ar": "", "title_en": "Autonomous control of a mobile robot with incremental deep learning neural networks", "author": "Isabella Glöde", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "deep-learning", "research_subdomain": "ai-industry", "categories": "deep-learning; robotics; ai-applications; ai-industry", "keywords": "", "abstract": "Over the last few years autonomous driving had an increasingly strong impact on the automotive industry. This created an increased need for artificial intelligence algo- rithms which allow for computers to make human-like decisions. However, a compro- mise between the computational power drawn by these algorithms and their subsequent performance must be found to fulfil production requirements. In this thesis incremental deep learning strategies are used for the control of a mobile robot such as a four wheel steering vehicle. This strategy is similar to the human approach of learning. In many small steps the vehicle learns to achieve a specific goal. The usage of incremental training leads to growing knowledge-base within the system. It also provides the opportunity to use older training achievements to improve the system, when more training data is available. To demonstrate the capabilities of such an algorithm, two different models have been formulated. First, a more simple model with counter wheel steering, and second, a more complex, nonlinear model with independent steering. These two models are trained incrementally to follow different types of trajectories. Therefore an algorithm was established to generate useful initial points. The incremental steps allow the robot to be positioned further and further away from the desired trajectory in the environ- ment. Afterwards, the effects of different trajectory types on model behaviour are investigated by over one thousand simulation runs. To do this, path planning for straight lines and circles are introduced. This work demonstrates that even simulations with simple network structures can have high performance.", "source_url": "https:\/\/openalex.org\/W3160493819", "pdf_url": "", "doi": "", "handle": "", "repository_name": "LA Referencia (Red Federada de Repositorios Institucionales de Publicaciones Científicas)", "source_record_id": "W3160493819", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1310, "title": "Sentiment Analysis Using Hybrid Machine Learning Technique", "title_ar": "", "title_en": "Sentiment Analysis Using Hybrid Machine Learning Technique", "author": "Nishant Kumar", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2016, "language": "en", "research_domain": "nlp", "research_subdomain": "sentiment-analysis", "categories": "machine-learning; deep-learning; nlp; expert-systems; sentiment-analysis", "keywords": "", "abstract": "It is observed that consumers often share their opinion, views or feeling about any term used on social network in the form of reviews, comments or feedback. Those feedbacks given by end users have a great impact for evolution of new version of any product. Due to this trend in social media in recent years, sentiment analysis has become an important concern for theoreticians and practitioners Moreover reviews are often written in natural language and are mostly unstructured. Thus, to obtain any meaningful information from these reviews, it needs to be processed. Due to large size of data it is impossible to process this information manually. Hence machine learning algorithms are considered for analysis. Since data are unstructured in nature, unsupervised machine learning algorithm can be helpful in solving this problem. But unsupervised methods have less accuracy; hence not acceptable. In this study, a hybrid machine learning approach is adopted to automatically find the requirements for next version of software. Also some reviews neither belong to positive cluster nor to negative. They mixed reaction or feeling about some topics. Those problem associated with NLP is solved using hybrid technique of the fuzzy c-means and ANN. Moreover in this study, different methods of unsupervised machine leaning algorithm are implemented and their results are compared with each other. The best outcome is used to train the neural network. By using this hybridization technique, accuracy gets increased. And in later stage, this technique is applied to find the new requirement of product.", "source_url": "https:\/\/openalex.org\/W2746211547", "pdf_url": "", "doi": "", "handle": "", "repository_name": "", "source_record_id": "W2746211547", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1312, "title": "Using machine learning for sentiment analysis of Slovene web commentaries", "title_ar": "", "title_en": "Using machine learning for sentiment analysis of Slovene web commentaries", "author": "Klemen Kadunc", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2016, "language": "en", "research_domain": "nlp", "research_subdomain": "sentiment-analysis", "categories": "machine-learning; nlp; predictive-analytics; sentiment-analysis", "keywords": "", "abstract": "The purpose of this work is to develop a tool for sentiment analysis of user comments. Several machine learning classifiers were tested and multinomial naive Bayes turned out to be the best predictor. We tried several preprocessing techniques, especially those for web texts. The classifier was improved with a Slovene sentiment lexicon, which is a list of words and set phrases with a positive and a negative connotation. An English sentiment lexicon was manually translated into Slovene. The analysed corpus of user comments was manually annotated by three annotators; its entries were selected from some of the most visited Slovene news portals. Both the lexicon and the annotated corpus of user comments are the main contributions of this work.", "source_url": "https:\/\/openalex.org\/W2409903970", "pdf_url": "", "doi": "", "handle": "", "repository_name": "", "source_record_id": "W2409903970", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1324, "title": "Novel symbolic and machine-learning approaches for text-based and multimodal sentiment analysis", "title_ar": "", "title_en": "Novel symbolic and machine-learning approaches for text-based and multimodal sentiment analysis", "author": "Soujanya Poria", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Stirling", "university_source": "repository_host", "college": "", "department": "", "country": "United Kingdom", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "nlp", "research_subdomain": "sentiment-analysis", "categories": "deep-learning; computer-vision; nlp; ai-applications; multimodal; sentiment-analysis; ai-education", "keywords": "", "abstract": "Emotions and sentiments play a crucial role in our everyday lives. They aid decision-making, learning, communication, and situation awareness in human-centric environments. Over the past two decades, researchers in artificial intelligence have been attempting to endow machines with cognitive capabilities to recognize, infer, interpret and express emotions and sentiments. All such efforts can be attributed to affective computing, an interdisciplinary field spanning computer science, psychology, social sciences and cognitive science. Sentiment analysis and emotion recognition has also become a new trend in social media, avidly helping users understand opinions being expressed on different platforms in the web. In this thesis, we focus on developing novel methods for text-based sentiment analysis. As an application of the developed methods, we employ them to improve multimodal polarity detection and emotion recognition. Specifically, we develop innovative text and visual-based sentiment-analysis engines and use them to improve the performance of multimodal sentiment analysis. We begin by discussing challenges involved in both text-based and multimodal sentiment analysis. Next, we present a number of novel techniques to address these challenges. In particular, in the context of concept-based sentiment analysis, a paradigm gaining increasing interest recently, it is important to identify concepts in text; accordingly, we design a syntaxbased concept-extraction engine. We then exploit the extracted concepts to develop conceptbased affective vector space which we term, EmoSenticSpace. We then use this for deep learning-based sentiment analysis, in combination with our novel linguistic pattern-based affective reasoning method termed sentiment flow. Finally, we integrate all our text-based techniques and combine them with a novel deep learning-based visual feature extractor for multimodal sentiment analysis and emotion recognition. Comparative experimental results using a range of benchmark datasets have demonstrated the effectiveness of the proposed approach.", "source_url": "https:\/\/openalex.org\/W2620364470", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Stirling Online Research Repository (University of Stirling)", "source_record_id": "W2620364470", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:01", "updated_at": "2026-11-15 16:53:01" }, { "id": 1326, "title": "Essays on Time Series Forecasting and Machine Learning", "title_ar": "", "title_en": "Essays on Time Series Forecasting and Machine Learning", "author": "Frank, Johannes", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; ai-applications; predictive-analytics; rnn; transformers; ai-industry", "keywords": "", "abstract": "This dissertation investigates how modern machine learning (ML) methods, with a particular focus on multi-task learning (MTL) frameworks, can enhance forecasts in financial and macroeconomic contexts. Chapter 1 outlines the limitations of traditional econometric approaches in environments characterized by nonlinearity, structural breaks, and mixed-frequency data, and motivates the use of ML models capable of exploiting shared information across related prediction tasks. Chapter 2 evaluates the performance of temporal fusion transformers (TFT) for forecasting realized volatility (RV) of US stocks. It shows that models with sector-level pooling significantly outperform asset-specific or fully pooled models, and that TFT surpasses traditional volatility models and common ML benchmarks, especially in volatile market phases. Chapter 3 extends this approach by introducing a two-dimensional MTL framework to jointly forecast RV across industry portfolios and forecast horizons. Using long short-term memory (LSTM) models with structured parameter sharing, the study finds that sharing information across both dimensions improves forecast accuracy, particularly at short horizons. Chapter 4 applies MTL to macroeconomic nowcasting using disaggregated US labor market and inflation data. Results show that MTL enhances nowcast accuracy compared to single-task and traditional models, with the optimal degree of information sharing depending on task-specific heterogeneity. Chapter 5 embeds MTL into a mixed frequency data sampling (MIDAS) setting by combining it with different MIDAS techniques to nowcast sectoral employment using high-frequency indicators. Nonlinear MTL models, particularly those based on extreme gradient boosting (XGBoost), outperform benchmarks and highlight the value of preserving intra-month dynamics. Overall, the dissertation demonstrates that combining MTL with flexible ML architectures offers a powerful framework for improving forecast quality and stability in complex economic settings.", "source_url": "https:\/\/openalex.org\/W7109568429", "pdf_url": "", "doi": "10.25593\/open-fau-2528", "handle": "", "repository_name": "", "source_record_id": "W7109568429", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:01", "updated_at": "2026-11-15 16:53:01" }, { "id": 1327, "title": "Forecasting euro area inflation using time series and machine learning techniques", "title_ar": "", "title_en": "Forecasting euro area inflation using time series and machine learning techniques", "author": "Vyshnivska, Ruslana", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Johannes Kepler Universitat Linz", "university_source": "repository_host", "college": "", "department": "", "country": "Austria", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "machine-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; predictive-analytics", "keywords": "", "abstract": "Machine learning models have been extensively used for forecasting inflation in individual countries, yet the number of studies focusing on the euro area is comparatively lower. The objective of this thesis is to compare the out-of-sample forecast performance of machine learning and traditional time series models on different forecast horizons for the euro area inflation. The focus is on core and headline inflation, as well as periods of high- and low-inflation. To this end, a comprehensive dataset was constructed to facilitate forecasting. The findings indicate that, in general, traditional time series models yield more accurate forecasts in the short and medium run. In contrast, machine learning models demonstrate higher forecast accuracy in the long run. This thesis also concludes that an increase in the number of predictors for machine learning models may reduce forecast performance when working with a small sample size. The findings further demonstrate that machine learning models may exhibit higher forecast accuracy for more volatile time series and during periods of high inflation. However, for forecasting core inflation in a low-inflation period, all models demonstrate similar performance, with no discernible pattern in forecast accuracy.", "source_url": "https:\/\/openalex.org\/W7113179973", "pdf_url": "", "doi": "", "handle": "", "repository_name": "University Library Linz repository (Johannes Kepler Universitat Linz)", "source_record_id": "W7113179973", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:01", "updated_at": "2026-11-15 16:53:01" }, { "id": 1329, "title": "Hybrid Machine Learning and Time-Series Modelling to Forecast Product Demand: A CRISP-DM Approach", "title_ar": "", "title_en": "Hybrid Machine Learning and Time-Series Modelling to Forecast Product Demand: A CRISP-DM Approach", "author": "Niklas Virtanen", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Vaasa", "university_source": "repository_host", "college": "", "department": "", "country": "Finland", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "nlp", "research_subdomain": "sentiment-analysis", "categories": "machine-learning; deep-learning; nlp; responsible-ai; ai-applications; predictive-analytics; sentiment-analysis; xai; ai-industry", "keywords": "", "abstract": "Efficient demand forecasting can yield significant competitive advantage for companies in today’s market, making it a crucial process in supply chain management. By exploring and implementing modern demand forecasting models, encompassing time-series, machine learning, deep learning and hybrid approaches, companies can gain benefits in important business areas such as strategic planning, operational efficiency and customer service. This research is done in collaboration with a case company, and it aims to identify and develop a product demand forecasting model that improves forecasting accuracy compared against a baseline approach and to identify and analyze key demand drivers that influence demand. To meet these goals, an industry standard CRISP-DM framework is utilized to develop and compare multiple time-series, machine learning and their hybrid combination models to find the best performing forecasting solution. Key demand drivers are identified and analyzed by combining SHAP values with statistical analysis. The results of this study indicate that a hybrid demand forecasting model combining Holt-Winters Exponential Smoothing and Elastic Net Regression outperformed the other models in the comparison and improved the forecasting accuracy significantly compared to the baseline approach. The results also suggest that product demand is driven by trend and seasonal factors, short-term sales pricing and momentum as well as macroeconomic leading indicators such as industry sentiment, interest rates and euro strength. Despite the significant improvement in forecasting accuracy and practical insights through the identified demand drivers, research limitations such as limited amount of training data, small test set and lack of true causal analysis of the relationships between features and demand are acknowledged to emphasize the need for future research. The development and evaluation of the HW-Elastic Net model contributes to current research by highlighting the importance of exploring hybrid models in demand forecasting, especially on small and noisy real-world datasets.", "source_url": "https:\/\/openalex.org\/W7198431739", "pdf_url": "https:\/\/osuva.uwasa.fi\/bitstreams\/6769593d-1bc1-41d8-9aac-273cac1b970a\/download", "doi": "", "handle": "", "repository_name": "Osuva (University of Vaasa)", "source_record_id": "W7198431739", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:01", "updated_at": "2026-11-15 16:53:01" }, { "id": 1337, "title": "Time Series Algorithms in Machine Learning - A Graph Approach to Multivariate Forecasting", "title_ar": "", "title_en": "Time Series Algorithms in Machine Learning - A Graph Approach to Multivariate Forecasting", "author": "Ryan Zhou", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Trent University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "deep-learning", "research_subdomain": "rnn", "categories": "machine-learning; deep-learning; ai-applications; predictive-analytics; rnn; gnn; ai-industry; ai-smart-cities", "keywords": "forecasting, graph neural network, LSTM, machine learning, neural network, time series", "abstract": "Forecasting future values of time series has long been a field with many and varied applications, from climate and weather forecasting to stock prediction and economic planning to the control of industrial processes. Many of these problems involve not only a single time series but many simultaneous series which may influence each other. This thesis provides methods based on machine learning of handling such problems. We first consider single time series with both single and multiple features. We review the algorithms and unique challenges involved in applying machine learning to time series. Many machine learning algorithms when used for regression are designed to produce a single output value for each timestamp of interest with no measure of confidence; however, evaluating the uncertainty of the predictions is an important component for practical forecasting. We therefore discuss methods of constructing uncertainty estimates in the form of prediction intervals for each prediction. Stability over long time horizons is also a concern for these algorithms as recursion is a common method used to generate predictions over long time intervals. To address this, we present methods of maintaining stability in the forecast even over large time horizons. These methods are applied to an electricity forecasting problem where we demonstrate the effectiveness for support vector machines, neural networks and gradient boosted trees. We next consider spatiotemporal problems, which consist of multiple interlinked time series, each of which may contain multiple features. We represent these problems using graphs, allowing us to learn relationships using graph neural networks. Existing methods of doing this generally make use of separate time and spatial (graph) layers, or simply replace operations in temporal layers with graph operations. We show that these approaches have difficulty learning relationships that contain time lags of several time steps. To address this, we propose a new layer inspired by the long-short term memory (LSTM) recurrent neural network which adds a distinct memory state dedicated to learning graph relationships while keeping the original memory state. This allows the model to consider temporally distant events at other nodes without affecting its ability to model long-term relationships at a single node. We show that this model is capable of learning the long-term patterns that existing models struggle with. We then apply this model to a number of real-world bike-share and traffic datasets where we observe improved performance when compared to other models with similar numbers of parameters. Author Keywords: forecasting, graph neural network, LSTM, machine learning, neural network, time series", "source_url": "https:\/\/openalex.org\/W7207033218", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Digital Collections (Trent University)", "source_record_id": "W7207033218", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:01", "updated_at": "2026-11-15 16:53:01" }, { "id": 1341, "title": "MACHINE LEARNING STRATEGIES TO PREDICT OIL FIELD PERFORMANCE AS TIME-SERIES FORECASTING", "title_ar": "", "title_en": "MACHINE LEARNING STRATEGIES TO PREDICT OIL FIELD PERFORMANCE AS TIME-SERIES FORECASTING", "author": "ISABEL FIGUEIRA DE ABREU GONCALVES", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "deep-learning", "research_subdomain": "rnn", "categories": "machine-learning; deep-learning; predictive-analytics; rnn", "keywords": "", "abstract": "Precisely forecasting oil field performance is essential in oil reservoir planning and management.Nevertheless, forecasting oil production is a complex nonlinear problem due to all geophysical and petrophysical properties that may result in different effects with a bit of change.Thus, all decisions to be made during an exploitation project must consider different efficient algorithms to simulate data, providing robust scenarios to lead to the best deductions.To reduce the uncertainty in the simulation process, recent studies have efficiently introduced machine learning algorithms for solving reservoir engineering problems since they can extract the maximum information from the dataset.This thesis proposes using two machine learning techniques to predict the daily oil production of an offshore reservoir.Initially, the oil rate production is considered a time series and is pre-processed and restructured to fit a supervised learning problem.The Random Forest model is used to forecast a one-time step, which is an extension of decision tree learning, widely used in regression and classification problems for supervised machine learning.Regardless, the restrictions of this approach lead us to a more robust model, the LSTM RNN's, which are proposed by several studies as a suitable deep learning technique for time series modeling.Various configurations of LSTM RNN's were constructed to implement single-step and multi-step oil rate forecasting and down-hole pressure was incorporated to the inputs.For testing the robustness of the proposed models, we use four different datasets, three of them synthetically generated and one from a public real dataset, the Volve oil field, as a case study to conduct the experiments.The results indicate that the Random Forest model could sufficiently estimate the one-time step of the oil field production, and LSTM could handle more inputs and adequately estimate multiple-time steps of oil production.", "source_url": "https:\/\/openalex.org\/W4381617672", "pdf_url": "https:\/\/www.maxwell.vrac.puc-rio.br\/62916\/62916.PDF", "doi": "10.17771\/pucrio.acad.62916", "handle": "", "repository_name": "", "source_record_id": "W4381617672", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:01", "updated_at": "2026-11-15 16:53:01" }, { "id": 1365, "title": "الذكاء الاصطناعي والأدوات المساعدة على الترجمة ما بين دعم كفـاءة المترجم وتهديد مستقبله المهني", "title_ar": "الذكاء الاصطناعي والأدوات المساعدة على الترجمة ما بين دعم كفـاءة المترجم وتهديد مستقبله المهني", "title_en": "", "author": "سمية عواد عيد عواد", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Algeria", "governorate": "", "publication_year": 2025, "language": "ar", "research_domain": "artificial-intelligence", "research_subdomain": "", "categories": "artificial-intelligence", "keywords": "", "abstract": "The purpose of this research entitled \"Artificial Intelligence and Translation Aids Between Supporting the Translator's Competence and Threatening his Professional Future\" is to solve the problem \"the reality of translation and the translator role under the presence of artificial intelligence and translation aids: is it a support for his competence or a threat to his professional future?\", to which we responded by abundantlying two theoretical chapters and a practical chapter. Thanks to this research, we have obtained several results, the most important of which is that despite the great challenge between machines and humans, they cannot overcome it because human capacities cannot be imitated in many aspects.", "source_url": "https:\/\/openalex.org\/W7135325394", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Dépôt Institutionnel de lUniversité de Tlemcen", "source_record_id": "W7135325394", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:01", "updated_at": "2026-11-15 16:53:01" } ] }