AI Glossary
Deep learning
deep learning, DL
Deep learning is a subfield of machine learning in which multi-layer neural networks automatically extract increasingly complex features from data. It is the technical foundation of today's language and generative models.
- It is a subfield of machine learning based on neural networks with many layers.
- Successive layers learn increasingly abstract features without anyone defining them by hand.
- It requires large datasets and substantial compute, typically on GPUs.
Deep learning is a narrower area within machine learning, built on neural networks made up of many layers. The word "deep" refers precisely to that number of layers: data passes through successive levels, and each one extracts features at a higher level of abstraction — from simple image edges to whole objects or the meaning of a sentence.
The difference from classical machine learning is a practical one. In the traditional approach, an engineer often hand-designs the features the model is meant to analyze. In deep learning, the network learns these representations itself from raw training data, which works exceptionally well for images, audio and text — but in return demands large datasets and substantial compute.
Deep learning is precisely what underpins today's AI: the transformer architecture, on which large language and generative models are based, is a deep neural network. The hierarchy of concepts is nested: artificial intelligence is the broadest, machine learning sits within it, and deep learning is a subset of machine learning. From a deployment standpoint, this means most of today's AI projects are, in essence, deep learning projects.
Related terms