AI glossary
Deep learning
Machine learning with neural networks that have many layers, able to learn complex patterns from raw data like text, images and sound.
“Deep” refers to the number of layers in the neural network. Each layer transforms the data a little, and together they can learn very abstract features — from pixels to edges to faces, or from letters to grammar to meaning.
Deep learning needs a lot of data and computing power, which is why it took off once GPUs and large datasets became available. Almost every modern AI system, including language models, is built with it.
Example: A speech recognition system learns from thousands of hours of transcribed audio. The first layers detect sounds, the next ones syllables and words, the last ones whole sentences.
In practice
- Almost everything we call AI today — translators, image recognition, chatbots — uses deep learning.
- Its weak spot is that it is hard to tell why it gives a particular answer; that is why interpretability is an active research field.
- To learn how it works inside, our resources list free courses from scratch.