The difference between AI, ML and DL
The relationship between AI, machine learning and deep learning, and when to use each.
What you will learn
- Drawing the relationship between the three concepts
- Knowing when deep learning is suitable and when it is not
Imagine three nested circles: artificial intelligence (AI) is the largest, inside it is machine learning (ML), and inside ML is deep learning. Deep learning uses multilayer neural networks that learn the features by themselves from raw data (pixels, texts, sounds).
In traditional machine learning we prepare the features ourselves (height, width, age, income) and then use algorithms such as trees and SVM. In deep learning the network learns the features layer by layer, but it needs more data and computation.
For small and medium tabular data, traditional machine learning algorithms are usually enough and are easier to interpret. For images, audio and long texts, deep learning usually wins.
Implementation steps with the tool
- Compare: a Teachable Machine model (deep transfer learning on images) with a Tree model in Orange on the Iris table.
Quick review question
Why don't we usually need a deep network to classify 150 flowers with four features?
