{ "exported_at": "2026-11-15 18:24:16", "count": 62, "records": [ { "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": 19, "title": "Fuzzy logic control of MPPT controller for PV systems", "title_ar": "", "title_en": "Fuzzy logic control of MPPT controller for PV systems", "author": "Mahmud Sasi", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Memorial University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "expert-systems", "research_subdomain": "ai-smart-cities", "categories": "expert-systems; ai-applications; ai-smart-cities", "keywords": "", "abstract": "This thesis presents a comparison between two methods to optimize the energy extraction in a photovoltaic (PV) power system. The maximum power of a PV module varies due to changing temperature, solar radiation, and load. To maximize efficiency, PV systems use a maximum power point tracker (MPPT) to constantly extract the highest power that can be produced by a solar panel and then deliver it to the load. The general structure of an MPPT system contains a DC-DC converter (an electronic device that converts a source of direct current DC from one voltage level to another) and a controller. The MPPT finds and maintains operations at the maximum power point using a tracking algorithm during variations in weather conditions. Many different algorithms of MPPT have been proposed and discussed in the literature, but most of these methods have disadvantages in terms of efficiency, accuracy, and flexibility. Because of the nonlinear behavior of PV module current-voltage characteristics and the nonlinearity of DC-DC converters due to switching, conventional controllers are unable to provide the best response, especially when dealing with wide parameter variations and line transients. The goal of this work is to design and implement a maximum power point tracker that uses a fuzzy logic control algorithm. Fuzzy logic naturally provides a superior controller for this type of nonlinear application. This method also benefits from the artificial intelligence approach for overcoming the complexity in modeling nonlinear systems. In order to succeed in this work, an MPPT system consisting of a PV module, a DC-DC converter, batteries, and a fuzzy logic controller is designed and simulated in Simulink. Analyses of buck, boost, and buck-boost converter characteristics are carried out to find the most suitable topology for the PV system used. An integrated model of the PV module with the identified converter and batteries is simulated in MATLAB to derive the expert knowledge needed to formulate and tune the fuzzy logic controller. The simulation results show that the fuzzy logic controller is able to obtain the desired outcomes and is ready to be applied to the hardware system. This entire research work aims to compare two types of controller-based MPPT techniques. Both MPPTs are based on the same topology of DC-DC converter and are applied with the same PV system specifications. That is, one of the MPPTs was kept at its original specifications and the other one was modified by changing the internal PIC 16F684 controller with an external Arduino Uno controller. Based on a MATLAB fuzzy logic design, the Arduino code was programmed and uploaded into an Arduino board by using Arduino software (IDE). The proposed method illustrates that the performance of MPPT is improved in terms of oscillations about the maximum power point, speed, and sensitivity to parameter variation. The results indicate that a significant amount of extra power can be extracted from a photovoltaic module by using a fuzzy logic-based maximum power point tracker in comparison with a PIC 16F684 controller-based maximum power tracker. Moreover, this gives improved efficiency for the operation of a PV power system, since batteries can be sufficiently charged and used during periods of low solar radiation.", "source_url": "https:\/\/openalex.org\/W2802625816", "pdf_url": "http:\/\/research.library.mun.ca\/12843\/1\/thesis.pdf", "doi": "", "handle": "", "repository_name": "Memorial University Research Repository (Memorial University)", "source_record_id": "W2802625816", "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": 48, "title": "Generating rhyming poetry using LSTM recurrent neural networks", "title_ar": "", "title_en": "Generating rhyming poetry using LSTM recurrent neural networks", "author": "Cole Peterson", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Victoria", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "deep-learning", "research_subdomain": "rnn", "categories": "deep-learning; rnn", "keywords": "", "abstract": "Current approaches to generating rhyming English poetry with a neural network involve constraining output to enforce the condition of rhyme. We investigate whether this approach is necessary, or if recurrent neural networks can learn rhyme patterns on their own. We compile a new dataset of amateur poetry which allows rhyme to be learned without external constraints because of the dataset’s size and high frequency of rhymes. We then evaluate models trained on the new dataset using a novel framework that automatically measures the system’s knowledge of poetic form and generalizability. We find that our trained model is able to generalize the pattern of rhyme, generate rhymes unseen in the training data, and also that the learned word embeddings for rhyming sets of words are linearly separable. Our model generates a couplet which rhymes 68.15% of the time; this is the first time that a recurrent neural network has been shown to generate rhyming poetry a high percentage of the time. Additionally, we show that crowd-source workers can only distinguish between our generated couplets and couplets from our dataset 63.3% of the time, indicating that our model generates poetry with coherency, semantic meaning, and fluency comparable to couplets written by humans.", "source_url": "https:\/\/openalex.org\/W3010625193", "pdf_url": "", "doi": "", "handle": "", "repository_name": "UVic’s Research and Learning Repository (University of Victoria)", "source_record_id": "W3010625193", "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": 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": 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": 115, "title": "Effective Explainable Artificial Intelligence using Visual Explanations in Images", "title_ar": "", "title_en": "Effective Explainable Artificial Intelligence using Visual Explanations in Images", "author": "Rami Ibrahim", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2022, "language": "en", "research_domain": "computer-vision", "research_subdomain": "cnn", "categories": "machine-learning; deep-learning; computer-vision; responsible-ai; predictive-analytics; cnn; xai", "keywords": "", "abstract": "Improved interpretations for CNN predictions.", "source_url": "https:\/\/openalex.org\/W4311681011", "pdf_url": "http:\/\/repository.library.carleton.ca\/downloads\/pc289k188", "doi": "10.22215\/etd\/2022-15159", "handle": "", "repository_name": "Carleton University's Institutional Repository (MacOdrum Library, Carleton University)", "source_record_id": "W4311681011", "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": 147, "title": "Class Imbalance and Time-To-Detection in the Performance Analysis of Machine Learning-Based Intrusion Detection Systems", "title_ar": "", "title_en": "Class Imbalance and Time-To-Detection in the Performance Analysis of Machine Learning-Based Intrusion Detection Systems", "author": "Shabanfar, Mohammad Pasha", "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": "computer-vision", "research_subdomain": "object-detection", "categories": "machine-learning; computer-vision; responsible-ai; ai-applications; object-detection; fairness; ai-cybersecurity; ai-industry; ai-smart-cities", "keywords": "", "abstract": "The increasing reliance on Industrial Control Systems (ICS) and Supervisory Control and Data Acquisition (SCADA) systems has raised critical concerns regarding their vulnerability to cyberattacks. While machine learning (ML) methods have emerged as effective tools for detecting such intrusions, their real-world applicability is challenged by two major issues: the imbalance in cybersecurity datasets and the limited focus on the time required to detect attacks—referred to as Time-To-Detection (TTD). To address these two issues and suggest better practices for ML-based IDS researchers, this thesis examines the gaps in the literature and, through two respective case studies, aims to suggest practices toward a more precise and practical performance assessment of ML-based intrusion detection systems (IDS). First, the thesis examines the class imbalance problem prevalent in popular Information Technology (IT) and Operational Technology (OT) cybersecurity datasets, where normal traffic often vastly outnumbers attack instances. This imbalance leads to biased model performance and inflated accuracy scores, which can over- or under-assess a model’s ability to identify the minority classes correctly. Through a case study with several ML models on a realistic dataset, we demonstrate how imbalanced classes should be considered in the performance evaluation, and how imbalance learning techniques like resampling should be properly utilized for robust performance of ML-based IDS. Second, this thesis examines TTD, a crucial but often understressed performance indicator that measures how promptly an ML-based detection system identifies the onset of an attack. In addition to traditional metrics that focus solely on classification performance in ML communities, this thesis proposed a TTD model based on real-world responsiveness in OT systems by defining vari ous stages of the detection process for ML researchers to quantify and measure temporal overheads accordingly. We also demonstrate how the proposed TTD model can be applied to OT datasets through a case study, thereby suggesting it as a best practice for a more comprehensive evaluation of ML-based intrusion detectors in practical OT use cases. Through the two studies above, the thesis offers a more comprehensive and practical approach to evaluating ML-based IDS. It demonstrates how thoughtful consideration and integration of class imbalance and detection timeliness in the development and assessment of ML-based IDS representation is essential to deploy trustworthy and efficient cybersecurity solutions in critical infrastructure systems.", "source_url": "https:\/\/openalex.org\/W7111482866", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Spectrum Research Repository (Concordia University)", "source_record_id": "W7111482866", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 198, "title": "Graph neural network for situation recognition", "title_ar": "", "title_en": "Graph neural network for situation recognition", "author": "Mohammed Suhail", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of British Columbia", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "deep-learning; computer-vision; transformers; gnn", "keywords": "", "abstract": "Understanding images beyond salient actions involves reasoning about scene con-text, objects, and the roles they play in the captured event. Situation recognition has recently been introduced as the task of jointly reasoning about the verbs (actions) and a set of semantic-role and entity (noun) pairs in the form of action frames. Labeling an image with an action frame requires an assignment of values (nouns) to the roles based on the observed image content. Among the inherent challenges are the rich conditional structured dependencies between the output role assignments and the overall semantic sparsity. In this work, we propose a novel mixture-kernel attention graph neural network (GNN) architecture designed to address these challenges. Our GNN enables dynamic graph structure during training and inference, through the use of a graph attention mechanism, and context-aware interactions between role pairs. It also alleviates semantic sparsity by representing graph kernels using a convex combination of learned basis. We illustrate the efficacy of our model and design choices by conducting experiments on imSitu benchmark dataset, with accuracy improvements of up to 10% over state-of-the-art", "source_url": "https:\/\/openalex.org\/W3003036658", "pdf_url": "", "doi": "10.14288\/1.0384601", "handle": "", "repository_name": "cIRcle (University of British Columbia)", "source_record_id": "W3003036658", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 211, "title": "Adaptive and Responsible Machine Learning: From Incremental Learning to Unlearning and Safe Generation", "title_ar": "", "title_en": "Adaptive and Responsible Machine Learning: From Incremental Learning to Unlearning and Safe Generation", "author": "Sayedmoslem Shokrolahi", "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": "computer-vision", "research_subdomain": "image-segmentation", "categories": "machine-learning; deep-learning; computer-vision; generative-ai; responsible-ai; ai-applications; multimodal; image-segmentation; diffusion; xai; fairness; ai-cybersecurity", "keywords": "", "abstract": "Modern machine learning (ML) systems are increasingly required to operate in dynamic environments, where models must adapt to new information, remove obsolete or sensitive knowledge, and satisfy fairness and safety requirements. This thesis addresses these challenges by developing mechanisms for controlling what a model learns, retains, and forgets. The first part studies incremental learning (IL) in data-free and class-incremental settings. To mitigate catastrophic forgetting and inter-task confusion, an embedding-based regularization strategy is proposed that separates current and previously learned class representations using stored class centroids. In addition, a generative replay mechanism reusing a previously trained model synthesizes representative samples of past tasks. A knowledge distillation formulation operating on embedding clusters is further developed to preserve intra-class and inter-class structures, enabling scalable IL without storing historical data. The second part investigates machine unlearning in discriminative models under strict data-free constraints. A new unlearning framework, termed MaxDiv, is proposed to remove the influence of designated training data by generating erasing samples that reverse the feature distribution of the data to be forgotten. The proposed method combines erasure-oriented objectives with representation-preserving losses to achieve effective forgetting while maintaining performance on remaining data. Experiments demonstrate improved forgetting guarantees, resistance to model inversion attacks, and superior preservation of retained knowledge compared to existing unlearning approaches. The third part extends machine unlearning to generative models, where the goal is often to control undesirable behaviors such as bias, unsafe content, and ethical misalignment. Following the responsible AI paradigm, this thesis studies responsible image generation in text-to-image (T2I) diffusion models (DMs) as a representation-level intervention problem. We first focus on U-Net-based DMs, where precise concept vectors are learned in the bottleneck layer to generate responsible images that preserve foreground and background content. We then demonstrate that multi-modal attention heads in ViT-based DMs exhibit interpretable and near-linear properties analogous to the bottleneck layer in U-Net-based architectures. Building on this observation, we introduce an architecture-agnostic framework for learning interpretable concept vectors that enable controllable, disentangled, and human-interpretable manipulation of internal representations across U-Net- and ViT-based diffusion backbones. Overall, this thesis provides a unified framework for adaptive and responsible ML.", "source_url": "https:\/\/openalex.org\/W7207902332", "pdf_url": "", "doi": "", "handle": "", "repository_name": "QSpace (Queen's University Library)", "source_record_id": "W7207902332", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 213, "title": "Trustworthy AI Under Distribution Shifts: Enhancing Generalization and Fairness in Deep Learning", "title_ar": "", "title_en": "Trustworthy AI Under Distribution Shifts: Enhancing Generalization and Fairness in Deep Learning", "author": "Gezheng Xu", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Western University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "medical-ai", "research_subdomain": "fairness", "categories": "machine-learning; deep-learning; computer-vision; generative-ai; medical-ai; responsible-ai; transfer-learning; predictive-analytics; optimization; fairness", "keywords": "", "abstract": "Recent advances in machine learning and deep learning have achieved remarkable success across a wide range of applications, yet they largely rely on data-driven paradigms that assume the training and test data share similar distributions. In real-world scenarios, however, this assumption often fails. Distribution shifts between training and deployment environments or among population subgroups can significantly compromise the performance and reliability of AI systems. Moreover, privacy constraints frequently prevent direct access to source data, and ethical concerns surrounding fairness raise additional challenges for the trustworthy use of AI in sensitive domains such as healthcare and finance. These factors together highlight the urgent need for a deeper and more principled understanding of trustworthy AI under distribution shifts. This thesis examines the influence of distribution shifts along two complementary dimensions of trustworthy AI: generalizability and fairness. On one hand, we focus on the noisy and unreliable signals caused by distribution shift, which place learning in a weak-supervision regime and thereby limit AI systems’ generalizability. On the other hand, we focus on the distribution shifts that arise from structural data imbalance or scarcity rooted in historical and societal factors, which give rise to socially unfavorable model behaviors and, in turn, exacerbate fairness issues in the decision-making process. From the generalizability perspective, the thesis studies the domain adaptation problem under data privacy constraints, formulated as the Source-Free Domain Adaptation (SFDA) problem. SFDA is examined in both classification and regression tasks, with a focus on pseudo-label noise mitigation and uncertainty modeling, respectively. For the classification problem, a noise-robust loss–based noise-and-variance control method (NVC-LLN) is proposed to mitigate the unbounded label noise inherent in SFDA. For the regression task, a histogram-based uncertainty reconstruction framework, MERCI, is proposed to refine continuous supervision signals and align feature presentations under distribution shift. For the fairness problem, this thesis focuses on the multiple-subgroup situation, ranging from a single sensitive attribute with multiple subgroups to multiple sensitive attributes with intersectional bias, covering both unfairness discovery and mitigation. For subgroup discovery, a Bias-Guided Generative Network (BGGN) is proposed to identify underrepresented subgroups instead of relying on traditional search methods. For mitigation, a generalization error bound is derived under the PAC-Bayes framework, motivating a probabilistic-predictor-based bilevel optimization algorithm, FAMS, which achieves both fairness and accuracy across multiple subgroups under limited-data conditions. Collectively, these studies advance the development of responsible, adaptive, and human-centric AI systems. These systems are capable of performing reliably across diverse and dynamic environments.", "source_url": "https:\/\/openalex.org\/W7205536362", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Scholarship@Western (Western University)", "source_record_id": "W7205536362", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 214, "title": "Engineering Fairness in Machine Learning Systems", "title_ar": "", "title_en": "", "author": "Moses Openja", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "École Polytechnique de Montréal", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2026, "language": "fr", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; nlp; medical-ai; responsible-ai; self-semi-supervised; llm; fairness", "keywords": "", "abstract": "s systèmes d’apprentissage automatique (AA), y compris les modèles classiques, les réseaux de neurones profonds (DNN) et les grands modèles de langage (LLMs), sont de plus en plus déployés dans des domaines à enjeux élevés tels que la santé, le recrutement, la fi-nance et les politiques publiques, où des décisions injustes ou biaisées peuvent avoir des conséquences sociétales majeures. Bien que des progrès substantiels aient été réalisés en matière d’apprentissage équitable, la plupart des approches existantes considèrent l’équité comme une contrainte statique lors de l’entraînement et supposent des distributions de don-nées fixes, des attributs sensibles prédéfinis et une possibilité de réentraîner sans restriction. En pratique, toutefois, les systèmes d’apprentissage automatique évoluent en permanence : le biais peut être introduit lors du prétraitement des données et de la sélection des carac-téristiques, émerger au cours de l’entraînement à travers l’apprentissage des représentations, persister dans des modèles préentraînés, ou être amplifié après le déploiement en raison de changements de distribution. Il est donc essentiel de garantir que l’équité n’est pas compro-mise tout au long du développement et de l’évolution des systèmes d’AA—non seulement pour produire des systèmes logiciels de qualité, mais aussi pour la conformité réglementaire et l’audit. Cette thèse envisage l’équité comme une propriété de qualité au niveau logi-ciel des systèmes d’AA, qui doit être continuellement testée, diagnostiquée et corrigée tout au long du cycle de vie du système. S’inspirant de l’ingénierie logicielle et de l’assurance qualité, la thèse adopte une démarche proactive en matière d’équité, mettant l’accent sur la détection précoce et la correction à faible coût des composants générant des biais, tout en développant également des mécanismes de correction post-entraînement et post-déploiement spécifiques aux contextes où le réentraînement n’est pas faisable. Aux stades des données et de l’entraînement, la thèse propose une méthode systématique pour identifier les carac-téristiques du modèle susceptibles de produire des biais, même sans connaître à l’avance les attributs sensibles. Grâce à une analyse cause-à-effet, cette approche permet aux experts du domaine de raisonner sur les sources de biais et démontre empiriquement que corriger ou supprimer les caractéristiques moins importantes mais génératrices de biais peut améliorer significativement l’équité sans détériorer la performance prédictive. Pour les DNN entraînés, la thèse introduit FairFLRep, un cadre de localisation et de répa-ration des défauts sensible à l’équité qui considère les comportements injustes comme des défectuosités localisées au sein des représentations apprises. FairFLRep identifie les poids de neurones contribuant de manière disproportionnée aux disparités entre sous-groupes et applique des mises à jour ciblées et minimales des paramètres, inspirées par les techniques de correction logicielle issues de l’ingénierie logicielle traditionnelle. En outre, la thèse étudie l’équité dans les modèles transformateurs préentraînés au travers de deux études complé-mentaires. La première examine la manière dont les biais sociaux sont représentés en interne dans les transformateurs, à partir d’un ensemble de données de relations biaisées — des triplets codant des stéréotypes selon divers types de biais — et de techniques d’attribution neuronale. L’étude montre que la connaissance biaisée est souvent concentrée dans de petits sous-ensembles de neurones tout en étant aussi redondamment encodée à travers le réseau. Ces résultats établissent le biais comme un phénomène à la fois architectural et représen-tatif, plutôt qu’un simple artéfact lié aux données, et en évaluent l’impact sur des tâches d’ingénierie logicielle en aval. Sur la base de ces observations, la seconde étude propose un cadre de réparation rigoureux pour les transformateurs préentraînés. Elle introduit une lo-calisation neuronale contrastive et optimale selon Pareto afin d’identifier les neurones de biais de manière concentrée et fiable, puis applique une optimisation axée sur la justesse à l’aide de la méthode d’optimisation par essaim particulaire (PSO) pour les corriger. Les résultats démontrent que la réparation ciblée au niveau des neurones est efficace pour atténuer les biais et constitue une alternative légère et pratique au réentraînement global. Enfin, la thèse aborde la réparation de l’équité sous changement de distribution, un phénomène fréquent dans les déploiements réels, même lorsque les facteurs sources du décalage de données ou de biais sont inconnus ou impossibles à mesurer. Pour dépasser les limites des approches de réparation statiques comme FairFLRep, la thèse propose DShift-FRep, un cadre adaptatif intégrant un alignement des données, un finetuning léger à point de départ chaud, et une réparation ciblée de neurones via PSO.", "source_url": "https:\/\/openalex.org\/W7220748911", "pdf_url": "https:\/\/publications.polymtl.ca\/76114\/1\/2026_MosesOpenja.pdf", "doi": "", "handle": "", "repository_name": "PolyPublie (École Polytechnique de Montréal)", "source_record_id": "W7220748911", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 221, "title": "FairPath: A Human-in-the-Loop Framework for Adversarial Fairness Testing in Machine Learning Pipelines", "title_ar": "", "title_en": "FairPath: A Human-in-the-Loop Framework for Adversarial Fairness Testing in Machine Learning Pipelines", "author": "Matheus de Morais Leça", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Calgary", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "responsible-ai", "research_subdomain": "fairness", "categories": "machine-learning; responsible-ai; fairness", "keywords": "", "abstract": "This thesis investigates how fairness is addressed within machine learning pipelines and proposes a framework for adversarial fairness testing through a human-in-the-loop approach. Although several fairness metrics and mitigation techniques have been proposed, existing approaches often present technical barriers that limit collaboration between machine learning practitioners and non-technical stakeholders, hindering the identification and mitigation of fairness issues before deployment. To address this gap, this research introduces a framework integrating fairness assessment, bias mitigation, explanatory feedback, and decision support mechanisms to enable accessible and structured fairness evaluation. A tool named FairPath was developed and evaluated through quantitative benchmarking and qualitative heuristic evaluation against comparable approaches. This thesis addresses five research questions: RQ1: How is bias identified and measured in the software fairness literature? RQ2: What data improvement approaches are proposed to enhance fairness? RQ3: What effects do data improvement approaches have on bias mitigation and representativeness enhancement? RQ4: What ethical considerations are examined in research on data improvement approaches? RQ5: Can a fairness tool enable accessible bias mitigation while preserving model utility? By addressing these questions, this research aims to bridge fairness assessment and stakeholder collaboration, reducing barriers to effective bias identification and mitigation while supporting the development of transparent, accountable, and socially responsible machine learning systems.", "source_url": "https:\/\/openalex.org\/W7212152316", "pdf_url": "https:\/\/ucalgary.scholaris.ca\/bitstreams\/26cedc4f-1a2c-494a-9087-abab61e79183\/download", "doi": "10.11575\/prism\/51959", "handle": "", "repository_name": "Libraries and Cultural Resources (University of Calgary)", "source_record_id": "W7212152316", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 223, "title": "Enhancing Fairness in Supervised Machine Learning", "title_ar": "", "title_en": "Enhancing Fairness in Supervised Machine Learning", "author": "Bita Omidi", "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": "medical-ai", "research_subdomain": "fairness", "categories": "machine-learning; deep-learning; medical-ai; responsible-ai; predictive-analytics; fairness", "keywords": "", "abstract": "The increasing influence of machine learning algorithms and artificial intelligence on the high-impact domains of decision-making has led to an increasing concern for the ethical and legal challenges posed by sensitive data-driven systems. Machine learning can identify the statistical patterns in the historically collected big data generated by a huge number of instances that might be affected by human and structural biases. ML algorithms have the potential to amplify these inequities. Lately, there have been several attempts to reduce bias in artificial intelligence in order to maintain fairness in machine learning projects. These methods fall under three categories of pre-processing, in-processing, and post-processing techniques. There are at least 21 notations of fairness in the recent literature, which not only provide different measurement methods of fairness but also lead to completely different concepts. It is worth mentioning that, it is impossible to satisfy all of the definitions of fairness at the same time and some of them are incompatible with each other. As a result, it is important to choose a fairness definition that need to be satisfied according to the context that we are working on. The current study investigates some of the most common definitions and metrics for fairness introduced by researchers to compare three of the proposed de-biasing techniques regarding their effects on the performance and fairness measures through empirical experiments on four different datasets. The de-biasing methods include the “Reweighing Algorithm”, “Adversarial De-biasing Method”, the “Reject Option Classification Method” performed on the classification tasks of “Survival of patients with heart failure”(Heart Failure Dataset), “Prediction of hospital readmission among diabetes patients” (Diabetes Dataset), “Credit classification of bank account holders” (German Credit Dataset), and “The COVID19 related anxiety level classification of Canadians” (CAMH Dataset). Findings show that the adversarial de-biasing in-processing method can be the best technique for mitigating bias working with the deep learning classifiers when we are capable of changing the classification process. This method has not led to a considerable reduction of accuracy except for the CAMH dataset. The “Reject Option Classification” which is a post-processing method, causes the most deterioration of prediction accuracy in all datasets. On the other hand, this method has the best performance in alleviating the bias generated through the classification process. The “Reweighing Algorithm” as a pre-processing technique does not cause a considerable reduction in the accuracy and is capable of reducing bias in classification tasks, although its performance is not as strong as the Reject Option classifier.", "source_url": "https:\/\/openalex.org\/W3196886932", "pdf_url": "https:\/\/ruor.uottawa.ca\/bitstreams\/a217da9e-f2fa-421a-b062-ba76c25339ab\/download", "doi": "10.20381\/ruor-26258", "handle": "", "repository_name": "uO Research (University of Ottawa)", "source_record_id": "W3196886932", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 227, "title": "\"Ethics, Governance, and Bias Mitigation in AI-Enabled loT Applications: A Focus on Transparency and Fairness", "title_ar": "", "title_en": "\"Ethics, Governance, and Bias Mitigation in AI-Enabled loT Applications: A Focus on Transparency and Fairness", "author": "Nadine Y. Fares", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "York University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "responsible-ai", "research_subdomain": "xai", "categories": "responsible-ai; ai-applications; xai; fairness; ai-industry; ai-smart-cities", "keywords": "", "abstract": "In the development of smart cities, the integration of Internet of Things (IoT) and Artificial Intelligence (AI) technologies promises to address urban challenges yet raises significant security and ethical concerns. This study emphasizes the need for robust governance frameworks to ensure the ethical deployment of these technologies, focusing on the roles of data privacy, security, and bias mitigation. It highlights the importance of Explainable AI (XAI) techniques, like LIME and SHAP, for enhancing transparency and accountability in AI decision-making processes. Through a detailed examination of existing governance frameworks and the introduction of a bias detection tool, this work aims to embed ethical considerations into the fabric of AI-driven systems in smart cities. The discussion extends to exploring common domains and AI implementations across different industries, offering insights into comprehending and ensuring the responsible use of AI and IoT technology. This lays the groundwork for future research in this developing area.", "source_url": "https:\/\/openalex.org\/W7171271463", "pdf_url": "https:\/\/yorkspace.library.yorku.ca\/bitstreams\/21897da7-867b-41da-9673-7e89552b2069\/download", "doi": "", "handle": "", "repository_name": "York University Digital Library (York University)", "source_record_id": "W7171271463", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 235, "title": "Forecasting the Next Winning Stock: A Comparative Analysis of Machine Learning Models", "title_ar": "", "title_en": "Forecasting the Next Winning Stock: A Comparative Analysis of Machine Learning Models", "author": "Fernandez Mendez, Blanca Elvira", "supervisor": "", "degree_type": "master", "degree_evidence": "نص السجل: «master»", "university": "York University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2025, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; predictive-analytics; transformers", "keywords": "", "abstract": "Stock price prediction is a common and complex problem due to the high volatility of financial markets. This master’s thesis presents a new approach to stock price forecasting by reformulating the problem as a multiclass classification task. The main objective is to predict which stock will yield the highest return the next day within a given set of features. To this end, various statistical and machine learning models are analyzed, with special emphasis on the Transformer model due to its relevance and alignment with the structure of this work. The present study proposes a novel idea to address the problem. Its contributions stand out in an initial exploratory analysis of model performance, as well as in risk minimization in investments, enabling portfolio diversification thanks to the Transformer model.", "source_url": "https:\/\/openalex.org\/W7112810442", "pdf_url": "", "doi": "", "handle": "", "repository_name": "York University Digital Library (York University)", "source_record_id": "W7112810442", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 236, "title": "Efficiency Advances in Moment Tensor Potentials for Machine-Learning Atomistic Modelling", "title_ar": "", "title_en": "Efficiency Advances in Moment Tensor Potentials for Machine-Learning Atomistic Modelling", "author": "Zijian Meng", "supervisor": "", "degree_type": "master", "degree_evidence": "نص السجل: «Master»", "university": "Queen's University Library", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2026, "language": "en", "research_domain": "machine-learning", "research_subdomain": "ai-education", "categories": "machine-learning; ai-applications; ai-education; ai-smart-cities", "keywords": "", "abstract": "The study of nuclear materials is strongly complemented by computational materials science (CMS), which provides an alternative means of investigating radiation-induced phenomena, defect evolution, and thermophysical properties—all of which can be hazardous, technically challenging, or impractical to study using traditional experimental methods. Within CMS, a central focus is the accurate atomic-level description of potential energy surfaces and atomic forces, which enables a wide range of simulations. These descriptions, known as interatomic potentials, have recently seen significant advancement through machine learning models. Co-supervised by the Departments of Mechanical and Materials Engineering, and Electrical and Computer Engineering, this Master’s thesis applies the principles of CMS, high-performance computing, and machine learning to advance the computational efficiency of Machine Learning Interatomic Potentials (MLIPs). While emphasis is placed on a specific MLIP, the Moment Tensor Potential (MTP), the methodologies developed herein can be extended for more general use. This thesis presents three contributions, each associated with a manuscript that has been published in, submitted to, or is under internal review before submission to a peer-reviewed journal. The first contribution introduces an efficient small-cell active learning protocol, achieving an order-of-magnitude reduction in training costs while maintaining high fidelity. The second contribution enhances the scalability of the MTP through an improved CPU implementation and a new GPU implementation, enabling higher simulation rates and the modeling of million-atom systems. The third contribution proposes a novel post-training pruning algorithm that is both data-driven and cost-aware, optimizing the MTP’s algorithmic structure and achieving up to sevenfold speedups. Collectively, these contributions establish a more efficient and scalable MTP framework, expanding the scale of accurate atomistic simulations for nuclear materials and for computational materials research more broadly.", "source_url": "https:\/\/openalex.org\/W7206226763", "pdf_url": "", "doi": "", "handle": "", "repository_name": "QSpace (Queen's University Library)", "source_record_id": "W7206226763", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:47", "updated_at": "2026-11-15 16:52:47" }, { "id": 298, "title": "Large-Scale Traffic Flow Prediction Using Deep Learning in the Context of Smart Mobility", "title_ar": "", "title_en": "Large-Scale Traffic Flow Prediction Using Deep Learning in the Context of Smart Mobility", "author": "Arief Barkah Koesdwiady", "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": "generative-ai", "research_subdomain": "gan", "categories": "machine-learning; deep-learning; generative-ai; ai-applications; predictive-analytics; gan; ai-smart-cities", "keywords": "", "abstract": "Designing and developing a new generation of cities around the world (termed as smart cities) is fast becoming one of the ultimate solutions to overcome cities' problems such as population growth, pollution, energy crisis, and pressure demand on existing transportation infrastructure. One of the major aspects of a smart city is smart mobility. Smart mobility aims at improving transportation systems in several aspects: city logistics, info-mobility, and people-mobility. The emergence of the Internet of Car (IoC) phenomenon alongside with the development of Intelligent Transportation Systems (ITSs) opens some opportunities in improving the tra c management systems and assisting the travelers and authorities in their decision-making process. However, this has given rise to the generation of huge amount of data originated from human-device and device-device interaction. This is an opportunity and a challenge, and smart mobility will not meet its full potential unless valuable insights are extracted from these big data. Although the smart city environment and IoC allow for the generation and exchange of large amounts of data, there have not been yet well de ned and mature approaches for mining this wealth of information to bene t the drivers and traffic authorities. The main reason is most likely related to fundamental challenges in dealing with big data of various types and uncertain frequency coming from diverse sources. Mainly, the issues of types of data and uncertainty analysis in the predictions are indicated as the most challenging areas of study that have not been tackled yet. Important issues such as the nature of the data, i.e., stationary or non-stationary, and the prediction tasks, i.e., short-term or long-term, should also be taken into consideration. Based on this observation, a data-driven traffic flow prediction framework within the context of big data environment is proposed in this thesis. The main goal of this framework is to enhance the quality of traffic flow predictions, which can be used to assist travelers and traffic authorities in the decision-making process (whether for travel or management purposes). The proposed framework is focused around four main aspects that tackle major data-driven traffic flow prediction problems: the fusion of hard data for traffic flow prediction; the fusion of soft data for traffic flow prediction; prediction of non-stationary traffic flow; and prediction of multi-step traffic flow. All these aspects are investigated and formulated as computational based tools\/algorithms\/approaches adequately tailored to the nature of the data at hand. The first tool tackles the inherent big data problems and deals with the uncertainty in the prediction. It relies on the ability of deep learning approaches in handling huge amounts of data generated by a large-scale and complex transportation system with limited prior knowledge. Furthermore, motivated by the close correlation between road traffic and weather conditions, a novel deep-learning-based approach that predicts traffic flow by fusing the traffic history and weather data is proposed. The second tool fuses the streams of data (hard data) and event-based data (soft data) using Dempster Shafer Evidence Theory (DSET). One of the main features of the DSET is its ability to capture uncertainties in probabilities. Subsequently, an extension of DSET, namely Dempsters conditional rules for updating belief, is used to fuse traffic prediction beliefs coming from streams of data and event-based data sources. The third tool consists of a method to detect non-stationarities in the traffic flow and an algorithm to perform online adaptations of the tra c prediction model. The proposed detection approach is developed by monitoring the evolution of the spectral contents of the traffic flow. Furthermore, the approach is specfi cally developed to work in conjunction with state-of-the-art machine learning methods such as Deep Neural Network (DNN). By combining the power of frequency domain features and the known generalization capability and scalability of DNN in handling real-world data, it is expected that high prediction performances can be achieved. The last tool is developed to improve multi-step traffic flow prediction in the recursive and multi-output settings. In the recursive setting, an algorithm that augments the information about the current time-step is proposed. This algorithm is called Conditional Data as Demonstrator (C-DaD) and is an extension of an algorithm called Data as Demonstrator (DaD). Furthermore, in the multi-output setting, a novel approach of generating new history-future pairs of data that are aggregated with the original training data using Conditional Generative Adversarial Network (C-GAN) is developed. To demonstrate the capabilities of the proposed approaches, a series of experiments using arti cial and real-world data are conducted.", "source_url": "https:\/\/openalex.org\/W2802252725", "pdf_url": "", "doi": "", "handle": "", "repository_name": "UWSpace (University of Waterloo)", "source_record_id": "W2802252725", "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": 309, "title": "Security of Smart Cyber-Physical Grids: A Deep Learning Approach", "title_ar": "", "title_en": "Security of Smart Cyber-Physical Grids: A Deep Learning Approach", "author": "Jacob Sakhnini", "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": "deep-learning", "research_subdomain": "ai-cybersecurity", "categories": "deep-learning; ai-applications; ai-cybersecurity; ai-smart-cities", "keywords": "", "abstract": "Cyber physical systems are widely used in critical infrastructure; among the most notable applications is the smart cyber-physical grid. The smart grid technologies are accompanied with various advantages including more efficient power generation and increased integration of green energy sources. As such, many cities around the world are investing in smart cyber-physical grid technologies. The use of this technology, however, comes with great risk to cyber threats. Furthermore, current state of the art defense methods lack in robustness, scalability,and computational efficiency. This thesis presents a deep learning based solution for attack detection in cyber-physical systems, particularly in the case of the smart cyber-physical grid.The research methods implemented in this thesis focus on improving robustness, scalability,and computational efficiency of intelligent attack detection algorithms by presenting heuristic methods for feature extraction and a novel deep learning approach that proved robust to varying attack sparsity and data imbalance.", "source_url": "https:\/\/openalex.org\/W3040743421", "pdf_url": "", "doi": "", "handle": "", "repository_name": "The Atrium (University of Guelph)", "source_record_id": "W3040743421", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 346, "title": "Deep Learning for 3D Information Extraction from Indoor and Outdoor Point Clouds", "title_ar": "", "title_en": "Deep Learning for 3D Information Extraction from Indoor and Outdoor Point Clouds", "author": "Ying Li", "supervisor": "", "degree_type": "phd", "degree_evidence": "نص السجل: «doctoral dissertation»", "university": "University of Waterloo", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "deep-learning; computer-vision; object-detection; image-segmentation", "keywords": "", "abstract": "This thesis focuses on the challenges and opportunities that come with deep learning in the extraction of 3D information from point clouds. To achieve this, 3D information such as point-based or object-based attributes needs to be extracted from highly-accurate and information-rich 3D data, which are commonly collected by LiDAR or RGB-D cameras from real-world environments. Driven by the breakthroughs brought by deep learning techniques and the accessibility of reliable 3D datasets, 3D deep learning frameworks have been investigated with a string of empirical successes. However, two main challenges lead to the complexity of deep learning based per-point labeling and object detection in real scenes. First, the variation of sensing conditions and unconstrained environments result in unevenly distributed point clouds with various geometric patterns and incomplete shapes. Second, the irregular data format and the requirements for both accurate and efficient algorithms pose problems for deep learning models. To deal with the above two challenges, this doctoral dissertation mainly considers the following four features when constructing 3D deep models for point-based or object-based information extraction: (1) the exploration of geometric correlations between local points when defining convolution kernels, (2) the hierarchical local and global feature learning within an end-to-end trainable framework, (3) the relation feature learning from nearby objects, and (4) 2D image leveraging for 3D object detection from point clouds. Correspondingly, this doctoral thesis proposes a set of deep learning frameworks to deal with the 3D information extraction specific for scene segmentation and object detection from indoor and outdoor point clouds. Firstly, an end-to-end geometric graph convolution architecture on the graph representation of a point cloud is proposed for semantic scene segmentation. Secondly, a 3D proposal-based object detection framework is constructed to extract the geometric information of objects and relation features among proposals for bounding box reasoning. Thirdly, a 2D-driven approach is proposed to detect 3D objects from point clouds in indoor and outdoor scenes. Both semantic features from 2D images and the context information in 3D space are explicitly exploited to enhance the 3D detection performance. Qualitative and quantitative experiments compared with existing state-of-the-art models on indoor and outdoor datasets demonstrate the effectiveness of the proposed frameworks. A list of remaining challenges and future research issues that help to advance the development of deep learning approaches for the extraction of 3D information from point clouds are addressed at the end of this thesis.", "source_url": "https:\/\/openalex.org\/W3140945473", "pdf_url": "", "doi": "", "handle": "", "repository_name": "UWSpace (University of Waterloo)", "source_record_id": "W3140945473", "other_sources": "", "verification_status": "verified", "created_at": "2026-11-15 16:52:48", "updated_at": "2026-11-15 16:52:48" }, { "id": 388, "title": "Transformer Block Coupling and Self-Attention Flows for Token Evolution", "title_ar": "", "title_en": "Transformer Block Coupling and Self-Attention Flows for Token Evolution", "author": "Anton Pavlovych Sugolov", "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": "Transformers and Large Language Models have achieved incredible performance in deep learning, especially in natural language processing, and a precise understanding of the internal mechanisms driving their success is essential. We first provide an empirical analysis of structural characteristics of transformer blocks, by analyzing the trajectories of LLM token embeddings as they traverse the model depth. We linearize the system along the trajectories through their Jacobian matrices and uncover transformer block coupling in a multitude of LLMs, characterized by the coupling of their top singular vectors across tokens and depth. Our findings reveal that coupling positively with model performance, and that this relationship is stronger than with other hyperparameters such as parameter count, model depth, and embedding dimension. Further, we investigate recent theoretical perspectives that cast transformers as iterative dynamical systems which modify embeddings. We provide an analysis of self-attention as a vector field that acts on Gaussian mixtures, building on recent works that construct interaction functionals on measures with similar properties to self-attention. Collectively, these insights offer a novel perspective on token interactions in transformers, opening new directions for studying their mechanisms as well as improving training and generalization.", "source_url": "https:\/\/openalex.org\/W7132879793", "pdf_url": "", "doi": "", "handle": "", "repository_name": "TSpace (University of Toronto)", "source_record_id": "W7132879793", "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": 393, "title": "MAWT: Multi-Attention-Weight Transformers", "title_ar": "", "title_en": "MAWT: Multi-Attention-Weight Transformers", "author": "Enver Askin", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; nlp; predictive-analytics; transformers; machine-translation", "keywords": "", "abstract": "Transformers are machine learning models designed to learn and predict sequential and structured data, which are crucial to tasks such as neural machine translation and semantic parsing.They have become state-of-the-art engines for both of these tasks and much research in natural language processing is devoted to increasing their performance by introducing modifications to their architectures.In light of this trend, this thesis introduces a new Transformer architecture called MAWT: Multi-Attention-Weight Transformers in an attempt to increase the accuracy and variety of the acceptable predictions of a Transformer.It attempts to achieve this by training multiple weights per each Transformer attention head, which then are used to test the accuracy of the engine.This creates a new architecture under which the system produces a candidate set of outputs (instead of a singly output), along with a method for selecting from the candidate set.My proposal rests on the assumption --motivated by statistical considerations --that having a candidate set increases the probability of finding an exact match within the set.Upon testing, I observed that my system outperforms the regular transformer on 5\/6 benchmark neural machine translation and semantic parsing datasets, where engine performance is measured by exact match accuracy.Exact match accuracy demands syntactic identity between the output and the target.In order to investigate how well my new architecture generalizes to measures of semantic equivalence that don't also demand syntactic identity, I also recorded the BLEU scores on these datasets.The BLEU score is a measure of performance based on n-grams rather than exact symbolic match (i.e., how many contiguous sequence of n-many strings from the predicted output match the desired output).The results I report on the BLEU scores are more iii mixed, raising important questions that I highlight about the role of syntax in measures of semantic equivalence.", "source_url": "https:\/\/openalex.org\/W4362577036", "pdf_url": "http:\/\/repository.library.carleton.ca\/downloads\/g732d990n", "doi": "10.22215\/etd\/2023-15442", "handle": "", "repository_name": "Carleton University's Institutional Repository (MacOdrum Library, Carleton University)", "source_record_id": "W4362577036", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 394, "title": "Sparse Attention Mechanisms in Vision Transformers: from Spatiotemporal Transformers to Instance-Dependent Attention Patterns", "title_ar": "", "title_en": "Sparse Attention Mechanisms in Vision Transformers: from Spatiotemporal Transformers to Instance-Dependent Attention Patterns", "author": "Brendan William Duke", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Toronto", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "deep-learning", "research_subdomain": "transformers", "categories": "machine-learning; deep-learning; computer-vision; nlp; predictive-analytics; cnn; transformers; image-segmentation; video-analysis; speech-recognition", "keywords": "", "abstract": "Deep learning (DL) algorithms have produced superior results for a variety of vision, language, and speech processing tasks, but have proven to be power and compute-hungry. In the past, hardware advancements have met the growing need for compute. Yet in recent years the end of Dennard scaling and Moore's law makes hardware advancements unlikely to continue at the same fast pace. Hence innovative algorithmic solutions are needed to meet DL algorithms' ever-growing demand for compute. In this dissertation we investigated methods using sparsity and conditional computation to meet this compute demand. The two main components of this contribution involve applying structure and dynamic prediction of sparsity patterns by machine learning algorithms. The hypothesis was that structured and dynamic sparse computation improves latency and memory footprint compared to existing approaches with equal algorithm effectiveness. To show this, we built on the state of the art in sparse computations. This dissertation fills in previously unexplored areas of applying structured sparsity to videos, and of predicting per-instance sparsity that is unstructured. The purpose was to investigate using structured and per-example sparsity to design performant and effective deep learning algorithms. Existing literature had investigated sparsity in computer vision, but only in the context of either convolutional neural networks (CNNs) or Transformers applied to image processing. Given the increased computational complexity, this raised the question of how sparsity fares when applied to video processing. We investigated this question in work presented at the CVPR 2021 conference called Sparse Spatiotemporal Transformers for Video Object Segmentation (SSTVOS). SSTVOS looks at the highly computationally demanding problem of applying Transformers to video by using sparse attention blocks. The sparse attention blocks in SSTVOS use a fixed sparsity pattern that takes advantage of a priori knowledge of which visual features to attend to. However, using fixed sparsity patterns limits us to problems where we have this a priori knowledge. Therefore we further executed a research direction to learn sparsity patterns that are performant (low latency and memory usage) and effective (superior predictive power) in a method called Sparsifiner. Sparsifiner predicts per-example sparsity patterns as a more flexible generalization of the fixed sparsity patterns in SSTVOS. With the expectation that a machine learning model learns to predict sparsity patterns superior to any fixed pattern designed by hand, the idea for Sparsifiner was born. Sparsifiner improves over existing sparsity methods in Vision Transformers~(ViTs) by predicting per-instance sparsity patterns with a computationally-efficient connectivity predictor module. The method serves as empirical evidence of the effectiveness of per-instance sparsity patterns compared to fixed sparsity patterns. Sparsifiner shows a Pareto-optimal trade-off between FLOPs and top-1 accuracy on ImageNet when compared to contemporary structured-sparse methods. Together, both SSTVOS and Sparsifiner constitute a basis for future work in the direction of structured and dynamic sparsity in deep learning.", "source_url": "https:\/\/openalex.org\/W7132914290", "pdf_url": "", "doi": "", "handle": "", "repository_name": "TSpace (University of Toronto)", "source_record_id": "W7132914290", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:49", "updated_at": "2026-11-15 16:52:49" }, { "id": 439, "title": "Natural Language Processing for Slang", "title_ar": "", "title_en": "Natural Language Processing for Slang", "author": "Zhewei Sun", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Toronto", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "nlp", "research_subdomain": "llm", "categories": "machine-learning; nlp; generative-ai; self-semi-supervised; predictive-analytics; llm", "keywords": "", "abstract": "Slang is a common type of language that makes creative and highly flexible use of words. A basic problem that language users tackle is how to develop and interpret novel slang terms for communication in a community. This problem is relevant for natural language processing since new slang expressions often emerge in daily conversations and online social media. However, principled computational approaches to modeling slang are lacking, which presents key challenges to the effective natural language processing of slang. In this dissertation, I develop a computational framework that offers new methodologies for the automated generation, interpretation, and translation of English slang word usages, as well as for characterizing the principles in slang variation across language communities. My dissertation is organized into three main parts. The first part addresses the under-explored problem of slang semantic extension, namely how existing words in the lexicon take on new meanings in informal context. I develop a generative framework that combines contrastive learning with probabilistic models of semantic chaining to capture slang semantic extension. By leveraging dictionary-based resources of slang, I show how the learned semantic representations more accurately predict slang word choices compared to existing approaches that rely more exclusively on corpus data. The second part of my dissertation tackles the inverse problem of slang interpretation by applying these semantic representations to interpret and translate novel slang usages in natural text. I show how this approach provides better accuracy and sample efficiency in both slang interpretation and translation, in comparison to baseline contextualized language models. Finally, the third part of my dissertation investigates semantic variation of slang across different language communities focusing on a comparative study of US and UK. I show that models incorporating either communicative need or semantic chaining can predict the regional identity of slang usages. In summary, my dissertation contributes a principled framework for modeling the lexical semantics and usages of English slang, and it opens up future opportunities for the computational investigation and automated processing of informal language across a diverse set of communities and languages.", "source_url": "https:\/\/openalex.org\/W7133071557", "pdf_url": "https:\/\/utoronto.scholaris.ca\/bitstreams\/2f5619f5-5d4c-40a8-bf85-3e19e6c0bfa5\/download", "doi": "", "handle": "", "repository_name": "TSpace (University of Toronto)", "source_record_id": "W7133071557", "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": 507, "title": "Graph Neural Networks for Circuit Clustering and Partitioning", "title_ar": "", "title_en": "Graph Neural Networks for Circuit Clustering and Partitioning", "author": "A R Soroush", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Guelph", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "deep-learning", "research_subdomain": "gnn", "categories": "machine-learning; deep-learning; data-mining; predictive-analytics; gnn", "keywords": "", "abstract": "Circuit clustering and circuit partitioning are essential tasks in the Very Large Scale Integration (VLSI) design flow and are usually solved using algorithmic techniques. Graph Neural Networks (GNNs), a branch of artificial neural networks, leverage the predictive power of deep learning on graph-structured data. Given the circuits' graphical nature, GNNs are increasingly applied to various VLSI design tasks. This research investigates applying GNNs to solve circuit clustering and partitioning. Moreover, a hierarchical framework is suggested, linking the clustering GNN to the partitioning GNN, where information gained from clustering GNN enriches the initial dataset before being fed to the partitioning GNN. The proposed method motivates further exploration of preceding a clustering GNN to other GNNs addressing other VLSI layout design problems (e.g. placement), enhancing the node features. Furthermore, the proposed linkage between two distinct GNNs can be extended to other GNNs that solve consecutive tasks in VLSI system design.", "source_url": "https:\/\/openalex.org\/W7204615144", "pdf_url": "https:\/\/atrium.lib.uoguelph.ca\/bitstreams\/6fc219c1-974a-4906-86a9-059e815dcc21\/download", "doi": "", "handle": "", "repository_name": "The Atrium (University of Guelph)", "source_record_id": "W7204615144", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:50", "updated_at": "2026-11-15 16:52:50" }, { "id": 538, "title": "Intelligent Robotic Recycling of Flat Panel Displays", "title_ar": "", "title_en": "Intelligent Robotic Recycling of Flat Panel Displays", "author": "Adam Sanderson", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Waterloo", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2019, "language": "en", "research_domain": "computer-vision", "research_subdomain": "object-detection", "categories": "deep-learning; computer-vision; robotics; ai-applications; cnn; object-detection; image-segmentation; ai-industry", "keywords": "", "abstract": "As the population and prosperity continue to rise the demand for high-tech products is rapidly increasing. Displays are a huge part of this market with millions being created each year. We have a finite amount of resources on this planet and it is imperative that we properly dispose of our devices. In this thesis an intelligent robotic workcell for dismantling and recycling of flat panel displays is proposed and tested. This system utilizes industrial robots and open source tools to process a flat panel display (FPD) with the goal of removal of cold compact fluorescent tubes (CFLs). These are removed so the FPD can be shredded and properly recycled without the mercury present in the CFLs contaminating the materials or harming workers. This system utilizes many new and innovative techniques including deep learning algorithms for object detection and image segmentation. These deep learning techniques specifically Faster R-CNN for object detection and DeepLab for image segmentation have been shown to be extremely robust and capable in this application. The system was tested both with a real world system and with simulated geometry with a live vision feed. During simulation, the system was capable of processing flat panel FPDs in between 110 and 230 seconds per unit. It has been calculated using the simulation results, that this system can be profitable and has an approximate payback period of between 0.19 and 4.87 years depending on the material being fed into the system. The large range is due to the difference in value between monitor and TV style FPDs. The monitors have far less value than the TVs and are far more difficult to process.", "source_url": "https:\/\/openalex.org\/W2947070151", "pdf_url": "https:\/\/uwspace.uwaterloo.ca\/bitstreams\/c1b45110-475c-469e-ab35-3d38398d358d\/download", "doi": "", "handle": "", "repository_name": "UWSpace (University of Waterloo)", "source_record_id": "W2947070151", "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": 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": 605, "title": "Synthetic Ultrasound Video Generation with Generative Adversarial Networks", "title_ar": "", "title_en": "Synthetic Ultrasound Video Generation with Generative Adversarial Networks", "author": "Daniil Kulik", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "", "university_source": "", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; generative-ai; medical-ai; cnn; video-analysis; gan; medical-imaging", "keywords": "", "abstract": "Ultrasound is a non-invasive, radiation-free portable imaging modality that offers real-time diagnostics in different clinical settings.Ultrasound video analysis can benefit from the rise of AI-powered applications in healthcare.However, access to ultrasound data remains the main challenge for the development of state-of-the-art machine learning models.Synthetic data generation can provide various benefits to ultrasound imaging analysis.Namely, generating ultrasound videos with specific characteristics allows for better training and testing of machine learning models.This work proposes a generative adversarial network for conditional ultrasound video generation.We conduct a thorough quantitative and qualitative evaluation of the network.Additionally, we show the added value of using the synthetic video for data augmentation in a downstream task.Extensive experiments on the EchoNet-Dynamic dataset demonstrate that the proposed model achieves an FID score of 126 and an FVD score of 233 and can be used in data augmentation tasks in small data scenarios.Contents 7 Conclusion and Future Work 70 vi List of Tables 3.1 Prior work in ultrasound image and video generation with GANs. .5.1 Configuration of the image discriminator. . . . . . . . . . . . . . . .5.2 Configuration of the video discriminator. . . . . . . . . . . . . . . .5.3 Configuration of the transposed convolutional generator. . . . . . .5.4 Configuration of the subpixel generator. . . . . . .", "source_url": "https:\/\/openalex.org\/W4389205759", "pdf_url": "https:\/\/repository.library.carleton.ca\/downloads\/w0892c38h", "doi": "10.22215\/etd\/2023-15790", "handle": "", "repository_name": "Carleton University's Institutional Repository (MacOdrum Library, Carleton University)", "source_record_id": "W4389205759", "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": 635, "title": "Reinforcement Learning for Circuit Partitioning", "title_ar": "", "title_en": "Reinforcement Learning for Circuit Partitioning", "author": "Ka Chuen Cheng", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Guelph", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "reinforcement-learning", "research_subdomain": "", "categories": "reinforcement-learning", "keywords": "", "abstract": "In the field of Electronic Design Automation (EDA), dividing a complex circuit into smaller subcircuits is a critical step. This process, known as circuit partitioning, minimizes the connections needed between these subcircuits while dividing the overall design. This can simplify the problem, and further enhance the performance and efficiency of other EDA tasks, such as placement and routing. A popular method for circuit partitioning is the Fiduccia-Mattheyses-Sanchis (FMS) algorithm, which is a local search heuristic. While it is efficient, the FMS algorithm has two weaknesses: it relies heavily on a good initial solution and can become caught in suboptimal solutions. This research explores various methods to construct better initial solutions for circuit partitioning. As well, it investigates the application of reinforcement learning (RL) techniques to overcome the local minima problem inherent in the FMS algorithm, thereby achieving higher-quality partitions.", "source_url": "https:\/\/openalex.org\/W7204611933", "pdf_url": "https:\/\/atrium.lib.uoguelph.ca\/bitstreams\/58895cdc-70da-492a-bebc-98d210b65dc3\/download", "doi": "", "handle": "", "repository_name": "The Atrium (University of Guelph)", "source_record_id": "W7204611933", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:51", "updated_at": "2026-11-15 16:52:51" }, { "id": 642, "title": "Multi-Scale Deep Learning Frameworks for Trajectory Prediction, Scene Understanding, and Perception in Autonomous Driving", "title_ar": "", "title_en": "Multi-Scale Deep Learning Frameworks for Trajectory Prediction, Scene Understanding, and Perception in Autonomous Driving", "author": "Xingnan Zhou", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Concordia University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "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; xai; fairness; ai-smart-cities", "keywords": "", "abstract": "Reliable autonomous vehicle operation at urban intersections requires competence across the prediction–perception stack, yet current deep learning models underuse the structural priors that govern real-world driving. This thesis argues that incorporating structural priors at multiple scales improves the accuracy, interpretability, and robustness of autonomous driving models, through five interconnected studies. Chapter 3 introduces a Turn-Aware LSTM that encodes turning intent via cumulative heading change. On UAV footage from a signalised intersection in Châteauguay, Québec, it reduces Final Displacement Error by 15–20% for turning manoeuvres at a 3-second horizon. Chapter 4 establishes lane graph conditioning as an architecture-flexible inductive bias for trajectory prediction. A lightweight module (under 50,000 parameters), evaluated on 89,258 Waymo scenarios, reduces ADE by 9.3% at 3 s and, at 8 s, minADE by 26.6%, minFDE by 32.6%, and miss rate by 42.7%. Chapter 5 recovers real-world locations for anonymised scenarios by matching intersection topologies to OpenStreetMap via WayGraph's 48-dimensional star pattern descriptor. Across 56,797 scenarios it achieves 90% top-1 accuracy under synthetic OSM-to-OSM matching (33–55% under realistic noise), and a geographic audit reveals fourfold under-sampling of signalised intersections (4.3% vs. 17.8% city-wide). Chapter 6 develops a spatial attention visualisation framework that projects Transformer attention onto bird's-eye-view scenes. Applied to MTR-Lite, it identifies a tunnel vision pattern: failed predictions are associated with lower attention entropy (5.72 vs. 5.94 bits) and elevated self-attention; cyclist targets show higher miss rates (88.1% vs. 54.0%). Left-turn manoeuvres trigger a general agent-to-map attention shift (−19.9% vehicle attention, p < 0.001), suggesting possible VRU-related attention vulnerabilities, though the cyclist-specific interaction did not reach significance (p = 0.51, n = 17) and requires validation on larger samples. Chapter 7 shows that viewpoint diversity suppresses false positives via dual-camera LiDAR fusion with vehicle- and drone-mounted cameras. Across 2,600 CARLA frames and ten seeds, symmetric late fusion improves mAP by +0.92 percentage points (p = 0.001) and reduces false positives by 13% on a representative seed. These studies support the thesis argument across five scales—kinematic, road topology, geographic, attentional, and sensor geometry—and outline future research directions.", "source_url": "https:\/\/openalex.org\/W7221486811", "pdf_url": "https:\/\/spectrum.library.concordia.ca\/id\/eprint\/997687\/6\/Zhou_PhD_F2026.pdf", "doi": "", "handle": "", "repository_name": "Spectrum Research Repository (Concordia University)", "source_record_id": "W7221486811", "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": 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": 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": 721, "title": "Deep Learning For The Classification of Lung Diseases Using Chest X-Rays", "title_ar": "", "title_en": "Deep Learning For The Classification of Lung Diseases Using Chest X-Rays", "author": "Khamesi, Zahra", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Concordia University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2023, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "deep-learning; computer-vision; medical-ai; medical-imaging", "keywords": "", "abstract": "The discovery of X-rays marked a significant milestone in the field of medicine. One of the most common types of X-rays, the chest X-ray (CXR), allows doctors to examine an individual’s internal structure without surgery. Over the years, deep learning methods and algorithms have been developed to automate lung disease detection and identification. This paper introduces RADIA, a project that combines multiple deep learning techniques to identify abnormal areas and abnormal- ities in chest X-rays. RADIA builds upon previous studies conducted by the Stanford ML group, such as ChexNet and ChexPert. Our team utilized the ConvNeXt-Large, a deep learning convo- lutional model, implemented with a pre-trained ConvNext algorithm on the ImageNet database to classify various pathologies from public datasets like ChestX-ray14, CheXpert, MIMIC-CXR, PadChest, and VinDr-CXR, as well as a private dataset obtained from the Picture Archiving Com- munication System (PACS) at Verdun and Notre Dame Hospitals in Montreal in the collaboration with CIUSSS (Centre Integre Universitaire de Sante et de Services Sociaux du Centre-Sud-de-l’Ile- de-Montreal) and valuable consultants from the radiology team at Notre Dame Hospital contributed to the project’s success. Our team employed image enhancement and augmentation techniques to create various image versions. We used different and novel approaches to address the challenges, and the results were evaluated using metrics such as AUC, F1, and G-Means to analyze performance with imbalanced input data. It is essential to note that the project’s development extends beyond the creation of a web tool based on deep learning techniques. Our future plans involve building a decision helper that combines inference models and web tools to assist healthcare professionals.", "source_url": "https:\/\/openalex.org\/W7112380474", "pdf_url": "https:\/\/spectrum.library.concordia.ca\/id\/eprint\/993141\/7\/Khamesi_MCompSc_S2024.pdf", "doi": "", "handle": "", "repository_name": "Spectrum Research Repository (Concordia University)", "source_record_id": "W7112380474", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 748, "title": "A Real-Time and Automatic Ultrasound-Enhanced Multimodal Second Language Training System: A Deep Learning Approach", "title_ar": "", "title_en": "A Real-Time and Automatic Ultrasound-Enhanced Multimodal Second Language Training System: A Deep Learning Approach", "author": "Mozaffari Maaref", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Ottawa", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "nlp", "research_subdomain": "speech-recognition", "categories": "machine-learning; deep-learning; computer-vision; nlp; medical-ai; ai-applications; multimodal; video-analysis; speech-recognition; medical-imaging; ai-education", "keywords": "", "abstract": "The critical role of language pronunciation in communicative competence is significant, especially for second language learners. Despite renewed awareness of the importance of articulation, it remains a challenge for instructors to handle the pronunciation needs of language learners. There are relatively scarce pedagogical tools for pronunciation teaching and learning, such as inefficient, traditional pronunciation instructions like listening and repeating. Recently, electronic visual feedback (EVF) systems (e.g., medical ultrasound imaging) have been exploited in new approaches in such a way that they could be effectively incorporated in a range of teaching and learning contexts. Evaluation of ultrasound-enhanced methods for pronunciation training, such as multimodal methods, has asserted that visualizing articulator’s system as biofeedback to language learners might improve the efficiency of articulation learning. Despite the recent successful usage of multimodal techniques for pronunciation training, manual works and human manipulation are inevitable in many stages of those systems. Furthermore, recognizing tongue shape in noisy and low-contrast ultrasound images is a challenging job, especially for non-expert users in real-time applications. On the other hand, our user study revealed that users could not perceive the placement of their tongue inside the mouth comfortably just by watching pre-recorded videos. Machine learning is a subset of Artificial Intelligence (AI), where machines can learn by experiencing and acquiring skills without human involvement. Inspired by the functionality of the human brain, deep artificial neural networks learn from large amounts of data to perform a task repeatedly. Deep learning-based methods in many computer vision tasks have emerged as the dominant paradigm in recent years. Deep learning methods are powerful in automatic learning of a new job, while unlike traditional image processing methods, they are capable of dealing with many challenges such as object occlusion, transformation variant, and background artifacts. In this dissertation, we implemented a guided language pronunciation training system, benefits from the strengths of deep learning techniques. Our modular system attempts to provide a fully automatic and real-time language pronunciation training tool using ultrasound-enhanced augmented reality. Qualitatively and quantitatively assessments indicate an exceptional performance for our system in terms of flexibility, generalization, robustness, and autonomy outperformed previous techniques. Using our ultrasound-enhanced system, a language learner can observe her\/his tongue movements during real-time speech, superimposed on her\/his face automatically.", "source_url": "https:\/\/openalex.org\/W3022567045", "pdf_url": "", "doi": "10.20381\/ruor-24710", "handle": "", "repository_name": "uO Research (University of Ottawa)", "source_record_id": "W3022567045", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "id": 749, "title": "Adaptive Fusion Techniques for Effective Multimodal Deep Learning", "title_ar": "", "title_en": "Adaptive Fusion Techniques for Effective Multimodal Deep Learning", "author": "Gaurav Sahu", "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": "deep-learning", "research_subdomain": "transformers", "categories": "deep-learning; computer-vision; nlp; generative-ai; transformers; multimodal; video-analysis; sentiment-analysis; machine-translation; speech-recognition; gan", "keywords": "", "abstract": "Effective fusion of data from multiple modalities, such as video, speech, and text, is a challenging task due to the heterogeneous nature of multimodal data. In this work, we propose fusion techniques that aim to model context from different modalities effectively. Instead of defining a deterministic fusion operation, such as concatenation, for the network, we let the network decide “how” to combine given multimodal features more effectively. We propose two networks: 1) Auto-Fusion network, which aims to compress information from different modalities while preserving the context, and 2) GAN-Fusion, which regularizes the learned latent space given context from complementing modalities. A quantitative evaluation on the tasks of multimodal machine translation and emotion recognition suggests that our adaptive networks can better model context from other modalities than all existing methods, many of which employ massive transformer-based networks.", "source_url": "https:\/\/openalex.org\/W3084985854", "pdf_url": "https:\/\/uwspace.uwaterloo.ca\/bitstreams\/d8ed4daf-ffb2-4594-8566-a7a3085f4b99\/download", "doi": "", "handle": "", "repository_name": "UWSpace (University of Waterloo)", "source_record_id": "W3084985854", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:52", "updated_at": "2026-11-15 16:52:52" }, { "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": 812, "title": "Analysis of the Niching Particle Swarm Optimization Algorithm", "title_ar": "", "title_en": "Analysis of the Niching Particle Swarm Optimization Algorithm", "author": "Tyler Crane", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Brock University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2021, "language": "en", "research_domain": "machine-learning", "research_subdomain": "optimization", "categories": "machine-learning; deep-learning; optimization; multimodal", "keywords": "", "abstract": "Multimodal optimization (MMO) techniques have been researched and developed over the years to track multiple global optima concurrently. MMO algorithms extend traditional unimodal optimization algorithms by using search strategies built around forming niches for multiple possible solutions. NichePSO was one of the first approaches to utilize particle swarm optimization (PSO) for MMO problems, using several small subswarms of agents working concurrently to form niches within the search space. Despite its promising performance NichePSO does suffer from some problems, and very little research has been done to study and improve upon the algorithm over the years. A main goal of this thesis is to analyze the NichePSO algorithm, gaining insight into the strengths and weaknesses of the algorithm. Empirical analyses were performed to study the NichePSO’s ability to maintain niches within complex problem domains, as well as methods for improving the overall performance and effectiveness of the algorithm. Two variants of the NichePSO algorithm are proposed, and experimental results show that they both significantly improve the performance of the NichePSO algorithm across several benchmark functions.", "source_url": "https:\/\/openalex.org\/W3175652444", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Brock University Digital Repository (Brock University)", "source_record_id": "W3175652444", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 818, "title": "Genetic learning particle swarm optimization for task matching in grid environment", "title_ar": "", "title_en": "Genetic learning particle swarm optimization for task matching in grid environment", "author": "Eid Mohammad Albalawi", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Manitoba", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "machine-learning", "research_subdomain": "optimization", "categories": "machine-learning; optimization", "keywords": "", "abstract": "Many scientific disciplines require processing computational-intensive applications with vast quantities of data. Large scale systems, such as Grid, provide shared resources on geographically distributed heterogeneous computing systems to meet the computational demands of the applications. One of the fundamental problem computing with the Grid system is assigning tasks of various users to the resources efficiently such that the execution time of the application is minimized while Grid resources are utilized optimally. This is called the task matching problem, an NP-hard problem, the focus of this thesis. While there are many heuristics for the task matching problem, particle swarm optimization (PSO) has been one of the latest heuristics in the literature adopted as a solution to this problem. However, PSO has few drawbacks, such as premature convergence to local optima. To circumvent these problems, we incorporate a learning approach inspired by the genetic algorithm to guide the PSO algorithm. We propose three novel genetic learning PSO algorithms: modified genetic learning PSO (MGLPSO), hybrid genetic learning PSO (HGLPSO), and modified genetic learning PSO with adaptive mutation (μ-MGLPSO). The proposed algorithms have been designed for both single and multi-objective task matching problem for tasks with and without dependencies. The proposed techniques are evaluated on standard benchmarks with large data sets and compared against state-of-the-art algorithms.", "source_url": "https:\/\/openalex.org\/W3083480189", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Mspace (University of Manitoba)", "source_record_id": "W3083480189", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 826, "title": "Optimizing convolutional neural network parameters using genetic algorithm for breast cancer classification", "title_ar": "", "title_en": "Optimizing convolutional neural network parameters using genetic algorithm for breast cancer classification", "author": "Khatereh Davoudi", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Manitoba", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2020, "language": "en", "research_domain": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; medical-ai; optimization; cnn; medical-imaging", "keywords": "", "abstract": "Breast cancer, as the most-regularly diagnosed cancer in women, can be controlled effectively by early-stage tumour diagnosis. Clinical specialists use Computer-Aided Diagnosis (CAD) systems to help aid in their diagnosis, as accurate as possible. Deep learning techniques, such as Convolutional Neural Network (CNN), due to their classification capabilities, have been widely adopted in CAD systems. The parameters of the network, including the weights of the convolution filters, and the weights of the fully connected layers play a crucial role in classification accuracy. Back-propagation technique is the most frequently used approach for training CNN. However, this technique has some disadvantages, such as getting stuck in local minima. In this thesis, we propose to optimize the weights of the CNN using Genetic Algorithm (GA). The work consists of: designing a CNN model to facilitate the classification process, training the model using three different optimizer (mini-batch gradient descent, Adam, and GA), and evaluating the model through various experiments on BreakHis dataset. We show that the CNN model trained through GA performs as well as the Adam optimizer with a classification accuracy of 85%.", "source_url": "https:\/\/openalex.org\/W3048136822", "pdf_url": "https:\/\/mspace.lib.umanitoba.ca\/bitstreams\/050cf8f8-9e09-4432-8764-8ad5bbcae9e1\/download", "doi": "", "handle": "", "repository_name": "Mspace (University of Manitoba)", "source_record_id": "W3048136822", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:53", "updated_at": "2026-11-15 16:52:53" }, { "id": 847, "title": "Medical Image Segmentation by Deep Convolutional Neural Networks", "title_ar": "", "title_en": "Medical Image Segmentation by Deep Convolutional Neural Networks", "author": "Qingbo Kang", "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": "medical-ai", "research_subdomain": "medical-imaging", "categories": "machine-learning; deep-learning; computer-vision; medical-ai; predictive-analytics; cnn; image-segmentation; medical-imaging", "keywords": "", "abstract": "Medical image segmentation is a fundamental and critical step for medical image analysis. Due to the complexity and diversity of medical images, the segmentation of medical images continues to be a challenging problem. Recently, deep learning techniques, especially Convolution Neural Networks (CNNs) have received extensive research and achieve great success in many vision tasks. Specifically, with the advent of Fully Convolutional Networks (FCNs), automatic medical image segmentation based on FCNs is a promising research field. This thesis focuses on two medical image segmentation tasks: lung segmentation in chest X-ray images and nuclei segmentation in histopathological images. For the lung segmentation task, we investigate several FCNs that have been successful in semantic and medical image segmentation. We evaluate the performance of these different FCNs on three publicly available chest X-ray image datasets. For the nuclei segmentation task, since the challenges of this task are difficulty in segmenting the small, overlapping and touching nuclei, and limited ability of generalization to nuclei in different organs and tissue types, we propose a novel nuclei segmentation approach based on a two-stage learning framework and Deep Layer Aggregation (DLA). We convert the original binary segmentation task into a two-step task by adding nuclei-boundary prediction (3-classes) as an intermediate step. To solve our two-step task, we design a two-stage learning framework by stacking two U-Nets. The first stage estimates nuclei and their coarse boundaries while the second stage outputs the final fine-grained segmentation map. Furthermore, we also extend the U-Nets with DLA by iteratively merging features across different levels. We evaluate our proposed method on two public diverse nuclei datasets. The experimental results show that our proposed approach outperforms many standard segmentation architectures and recently proposed nuclei segmentation methods, and can be easily generalized across different cell types in various organs.", "source_url": "https:\/\/openalex.org\/W3044795341", "pdf_url": "https:\/\/spectrum.library.concordia.ca\/985550\/1\/Kang_MASc_F2019.pdf", "doi": "", "handle": "", "repository_name": "Spectrum Research Repository (Concordia University)", "source_record_id": "W3044795341", "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": 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": 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": 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": 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": 1158, "title": "Achieving Consumable Big Data Analytics by Distributing Data Mining Algorithms", "title_ar": "", "title_en": "Achieving Consumable Big Data Analytics by Distributing Data Mining Algorithms", "author": "Shady Khalifa", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Queen's University Library", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2017, "language": "en", "research_domain": "computer-vision", "research_subdomain": "data-mining", "categories": "machine-learning; computer-vision; data-mining; predictive-analytics", "keywords": "", "abstract": "Businesses look at Big Data as an opportunity to gain insights for improving their services. The derivation of such insights requires using different data mining techniques. Mature data mining tools like WEKA or R have been in development for years. They implement a large number of data mining algorithms and can support sophisticated Analytics. However, these mature tools are designed to run on a single machine making them unsuitable to handle Big Data. Using these tools requires data mining and statistics knowledge, and some of them, like R, are hard to learn. Businesses do not always have the technical skills required to carry on such Analytics. Even if they do, it is challenging to find a tool with the needed algorithms that supports distributed processing to handle the Big Data high arrival velocity and large volumes. The Businesses’ analytical requirements can be addressed by Consumable Big Data Analytics, that is, solutions that allow businesses to do Big Data Analytics themselves using their in-house expertise. In this work, we provide a Consumable Analytics solution to meet the businesses’ analytical needs. First, we conduct a survey of existing Analytics solutions to identify possible areas of improvement to provide Consumable Analytics. Second, instead of developing distributed data mining algorithms to handle Big Data, we develop the Data Mining Distribution (DMD) algorithm and the Label-Aware Disjoint Partitioning (LADP) algorithm to distribute the execution of all existing single-machine data mining algorithms without rewriting a single line of their code. This gives users the flexibility to use any available data mining library, have algorithms like Hoeffding Tree run 70% to 95% faster and achieve up to 18% increase in prediction accuracy. Third, we develop the free and open source QDrill solution to implement our DMD and LADP algorithms for distributed Analytics. QDrill implements our proposed Distributed Analytics Query Language (DAQL) interface that adds Analytics capabilities to the regular SQL syntax and allows integration with Business Intelligence (BI) tools. This allows businesses to use their in-house expertise to do Big Data Analytics using the spreadsheets and visualizations of their BI tools.", "source_url": "https:\/\/openalex.org\/W2608234461", "pdf_url": "", "doi": "", "handle": "", "repository_name": "QSpace (Queen's University Library)", "source_record_id": "W2608234461", "other_sources": "", "verification_status": "needs_review", "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": 1225, "title": "Becoming the Cyberattacker: Active Defense for the Electric Grid", "title_ar": "", "title_en": "Becoming the Cyberattacker: Active Defense for the Electric Grid", "author": "Amr Mohamed Saber Mohamed", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Toronto", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2024, "language": "en", "research_domain": "reinforcement-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; reinforcement-learning; ai-applications; predictive-analytics; ai-cybersecurity", "keywords": "", "abstract": "Understanding the threat landscape of cyber-physical systems is crucial for developing effective defenses against emerging, highly sophisticated cyber threats. This understanding demands two essential components: the ability to predict the sophisticated behaviors of threat actors or the capacity to engage with them to understand their tactics and strategies. Currently, we face challenges in both areas -- lacking effective methods to model sophisticated cyberattackers or engage with them in meaningful ways. This research addresses these challenges by introducing frameworks that leverage physics-aware artificial intelligence to model sophisticated threat actors. Additionally, the research contributes to the development of frameworks that enable intelligent interactions with real cyberattackers. Succinctly, this research makes key contributions in three research directions. First, it contributes a novel framework that leverages reinforcement learning for modeling attackers. Second, it contributes another innovative framework that leverages safety-critical control to create resilient honeypots capable of intelligently engaging with sophisticated cyber attackers. Third, the research contributes several novel defense approaches, including the use of synthetic attack data for data-driven threat detection and risk assessment, the utilization of synthetic attacker models to validate and test defenses, and the application of safety-critical control for threat mitigation and safe reinforcement learning for dynamic defense. The benefits arising from these contributions are extensive. In research, these contributions provide methods for automating vulnerability discovery and unveiling strategic cyber threats capable of exploiting them. This proactive comprehension of cyber risks facilitates the active design of secure and resilient systems against emerging cyber threats. Additionally, the frameworks offer an approach to test and validate defenses against realistic strategic attacks. In practical application, the defensive contributions comprise data-driven methods capable of monitoring cyber-physical systems and offering early threat detection, threat classification, and risk assessment to assist in effective cyber incident response. Furthermore, the contributions related to honeypots present methods for security teams to deploy intelligent honeypots that can lure, deceive, and interact with attackers, providing security teams with a means to gather intelligence about the threat landscape. This, in turn, enables a better understanding of attackers' resources and capabilities, facilitating improved anticipation and mitigation of threats. Collectively, these benefits converge into an active defense stance, empowering organizations to proactively safeguard against cyber threats. The frameworks developed in this research present a broad, innovative, and advanced approach to understanding, modeling, and ultimately defending against the ever-evolving landscape of cyber threats targeting cyber-physical systems. فهم مشهد التهديدات لأنظمة التحكم السيبرانية-الفيزيائية أمر بالغ الأهمية لتطوير دفاعات فعّالة ضد التهديدات السيبرانية الناشئة والمعقدة. يتطلب هذا الفهم عنصرين أساسيين: القدرة على التنبؤ بسلوك الجهات المهاجمة المتطورة أو القدرة على التفاعل معهم لفهم تكتيكاتهم واستراتيجياتهم. نواجه حالياً تحديات في كلا الجانبين، حيث نفتقر إلى أساليب فعالة لنمذجة المهاجمين السيبرانيين المتقدمين أو التفاعل معهم بطرق ذات مغزى. تتناول هذه الدراسة تلك التحديات من خلال تقديم أطر عمل تعتمد على الذكاء الاصطناعي الملمّ بالفيزياء لنمذجة الجهات المهاجمة المتقدمة. بالإضافة إلى ذلك، تساهم هذه الدراسة في تطوير أطر تمكّن من التفاعل الذكي مع المهاجمين السيبرانيين الحقيقيين. بإيجاز، تقدم هذه الدراسة ثلاث مساهمات رئيسية في ثلاثة اتجاهات بحثية. أولاً، تساهم في إطار عمل جديد يعتمد على التعلم المعزز لنمذجة المهاجمين. ثانياً، تقدم إطارًا مبتكرًا آخر يعتمد على التحكم في الأنظمة الحرجة من أجل إنشاء \"مصائد ذكية\" قادرة على التفاعل بذكاء مع المهاجمين السيبرانيين المتقدمين. ثالثاً، تساهم الدراسة في تطوير عدة أساليب دفاعية جديدة، بما في ذلك استخدام بيانات هجمات صناعية لاكتشاف التهديدات وتقييم المخاطر، وتطبيق نماذج مهاجمين صناعية لاختبار الدفاعات والتحقق منها، واستخدام التحكم في الأنظمة الحرجة لتخفيف التهديدات وتطبيق التعلم المعزز الآمن للدفاع الديناميكي. تتمثل الفوائد الناتجة عن هذه المساهمات في نطاق واسع. في مجال البحث، تقدم هذه المساهمات طرقًا لاكتشاف الثغرات تلقائيا وكشف التهديدات الاستراتيجية القادرة على استغلالها. هذا الفهم الاستباقي للمخاطر السيبرانية يمكّن من التصميم الفعّال لأنظمة آمنة وقابلة للتكيف ضد التهديدات السيبرانية الناشئة. علاوة على ذلك، تقدم الأطر المطورة أسلوبًا لاختبار الدفاعات والتحقق منها ضد هجمات واقعية واستراتيجية. في التطبيق العملي، تتضمن المساهمات الدفاعية أساليب تعتمد على البيانات يمكنها مراقبة الأنظمة السيبرانية-الفيزيائية وتقديم كشف مبكر عن التهديدات، وتصنيفها، وتقييم المخاطر للمساعدة في الاستجابة الفعّالة للحوادث السيبرانية.", "source_url": "https:\/\/openalex.org\/W7226042570", "pdf_url": "https:\/\/utoronto.scholaris.ca\/bitstreams\/5a5309c3-cc14-4742-9e49-043d82f40563\/download", "doi": "", "handle": "", "repository_name": "TSpace (University of Toronto)", "source_record_id": "W7226042570", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:52:59", "updated_at": "2026-11-15 16:52:59" }, { "id": 1248, "title": "Reinforcement Learning on Resource Bounded Systems", "title_ar": "", "title_en": "Reinforcement Learning on Resource Bounded Systems", "author": "Jaden B. Travnik", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "University of Alberta", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "deep-learning", "research_subdomain": "predictive-analytics", "categories": "machine-learning; deep-learning; reinforcement-learning; ai-applications; predictive-analytics; ai-industry", "keywords": "", "abstract": "Recent advancements in reinforcement learning have made the field interesting to academia and industry alike. Many of these advancements depend on deep learning as a means to approximate a value function or a policy. This dependency usually relies on high performance hardware (e.g., a graphics processing unit, GPU) and applications of deep learning are often limited to domains where a substantial amount of time is allowed for a prediction to be made or an action to be chosen. Although these criteria cover many important use cases, the growing popularity of the ``Internet of Things,'' wearable electronics, and the advancement of myoelectric prosthetic limbs presents a rapidly growing real-time domain of resource bounded systems that are not properly suited for deep learning yet could still benefit from the application of reinforcement learning. Furthermore, many of these systems are limited by the physical space they must occupy. This restricts the size of the hardware and thereby the computational and memory resources that can be used. Despite these restrictions, demand for prompt actions from receptive systems continues to grow. To address these problems, I first highlight the difficulties one is faced with when implementing reinforcement learning on a system which is deployed in an asynchronous environment and introduce a new metric of performance by measuring the time it takes for a system to react to an observed state of an asynchronous environment. Secondly, I develop a class of algorithms that addresses these issues by reordering the algorithmic components to minimize reaction time. Thirdly, to minimize both the time and memory necessary to compute function approximation, I introduce a novel linear function approximation method, selective Kanerva coding (SKC), that allows a reinforcement learning agent to perform behaviors reactively in real-time while using less memory and computation time than the standard linear approach of tile coding. I also show that SKC is less sensitive to the curse of dimensionality than tile coding making SKC a significant step towards accurately representing high dimensional data on resource bounded systems. Moreover, I show that SKC can make the inclusion of more sensory modalities more feasible, which can increase prediction accuracy when those modes of sensation are relevant to the prediction. Finally, I present an exploration of the meta-parameters of SKC and evaluate the performance of two different variations of SKC against the original formulation. These findings are imperative to the current state of the field of reinforcement learning as they form a challenging perspective that is contrary to the current direction of the field's focus on deep learning. I form this argument by emphasizing the impracticality of deep learning in domains of resource bounded systems deployed in real-time environments, establishing the limitations on available computation and memory of these systems, and address these issues by proposing new insights, algorithms, and representations.", "source_url": "https:\/\/openalex.org\/W2809564569", "pdf_url": "", "doi": "10.7939\/r39g5gv5s", "handle": "", "repository_name": "ERA: Education and Research Archive (University of Alberta)", "source_record_id": "W2809564569", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "id": 1307, "title": "Sentiment analysis on Twitter data using machine learning", "title_ar": "", "title_en": "Sentiment analysis on Twitter data using machine learning", "author": "Ravikumar Patel", "supervisor": "", "degree_type": "unspecified", "degree_evidence": "", "university": "Laurentian University", "university_source": "repository_host", "college": "", "department": "", "country": "Canada", "governorate": "", "publication_year": 2018, "language": "en", "research_domain": "nlp", "research_subdomain": "sentiment-analysis", "categories": "machine-learning; computer-vision; nlp; data-mining; sentiment-analysis; speech-recognition", "keywords": "", "abstract": "In the world of social media people are more responsive towards product or certain events that are currently occurring. This response given by the user is in form of raw textual data (Semi Structured Data) in different languages and terms, which contains noise in data as well as critical information that encourage the analyst to discover knowledge and pattern from the dataset available. This is useful for decision making and taking strategic decision for the future market. To discover this unknown information from the linguistic data Natural Language Processing (NLP) and Data Mining techniques are most focused research terms used for sentiment analysis. In the derived approach the analysis on Twitter data to detect sentiment of the people throughout the world using machine learning techniques. Here the data set available for research is from Twitter for world cup Soccer 2014, held in Brazil. During this period, many people had given their opinion, emotion and attitude about the game, promotion, players. By filtering and analyzing the data using natural language processing techniques, and sentiment polarity has been calculated based on the emotion word detected in the user tweets. The data set is normalized to be used by machine learning algorithm and prepared using natural language processing techniques like Word Tokenization, Stemming and lemmatization, POS (Part of speech) Tagger, NER (Name Entity recognition) and parser to extract emotions for the textual data from each tweet. This approach is implemented using Python programming language and Natural Language Toolkit (NLTK), which is openly available for academic as well as for research purpose. Derived algorithm extracts emotional words using WordNet with its POS (Part-of-Speech) for the word in a sentence that has a meaning in current context, and is assigned sentiment polarity using ‘SentWordNet’ Dictionary or using lexicon based method. The resultant polarity assigned is further analyzed using Naïve Bayes and SVM (support vector Machine) machine learning algorithm and visualized data on WEKA platform. Finally, the goal is to compare both the results of implementation and prove the best approach for sentiment analysis on social media for semi structured data.", "source_url": "https:\/\/openalex.org\/W7071399435", "pdf_url": "", "doi": "", "handle": "", "repository_name": "Lu Zone Ul (Laurentian University)", "source_record_id": "W7071399435", "other_sources": "", "verification_status": "needs_review", "created_at": "2026-11-15 16:53:00", "updated_at": "2026-11-15 16:53:00" }, { "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" } ] }