Skip to content
LiveNext 02:05:21

Definition

What is Deep learning?

Deep learning is Deep learning is a kind of machine learning that uses neural networks with many stacked layers to learn patterns directly from raw data such as text, images and audio.

What deep learning is

Deep learning is a branch of machine learning built on neural networks with many layers. The "deep" refers to the number of layers, not to depth of understanding.

Earlier machine learning often needed people to hand-pick the features that mattered, such as the edges in an image or the word counts in an email. Deep learning skips that step. Given enough examples, the network works out its own features, layer by layer.

How it works in plain terms

Picture an assembly line for recognizing a face. The first station spots tiny edges. The next combines edges into curves and corners. The next builds eyes and noses from those. The last decides whose face it is. Each station is a layer, and none was told what to look for. Training taught each one what to pass along.

Training adjusts millions or billions of parameters using gradient descent, usually on GPUs because the math is a huge pile of matrix multiplication that those chips do quickly.

Common architectures

Why it matters to you

Deep learning is why AI jumped forward in the last decade. Speech recognition, translation, image generation and chatbots all rely on it. It also explains the costs. These models need enormous training data and compute, and they are hard to inspect. Even their builders often cannot say exactly why a given answer came out, which is why evals and testing matter so much.

An example

A hospital trains a deep learning model on many thousands of scans labeled by radiologists. Without being told what a tumor looks like, the network learns visual patterns that separate healthy tissue from suspicious tissue. It then flags new scans for a doctor to review.

How it relates to neighboring terms

Deep learning sits inside machine learning, which sits inside AI. A large language model is a deep learning model trained on text. Fine-tuning takes an already trained deep learning model and adapts it to a narrower job.

Questions people ask

What is the difference between machine learning and deep learning?

Deep learning is a subset of machine learning that uses neural networks with many layers and learns its own features from raw data, while other machine learning methods often rely on hand-chosen features.