Neural Network

A model built from layers of simple weighted units, trained by adjusting those weights until the output matches the target.

Machine Learning Beginner 1 min read Start here

Also known as: Artificial Neural Network, ANN

Definition

A neural network is a stack of layers, each applying a linear transformation followed by a non-linear function. Individually these units are trivial; composed in depth, they can approximate remarkably complicated functions.

Training means comparing the network's output to a target, computing how wrong it was, and nudging every weight in the direction that reduces the error. Repeat across a large dataset and useful structure emerges in the intermediate layers without anyone specifying it.

The biological analogy that gave the field its name is loose enough to be more misleading than helpful — these are differentiable function approximators, not simulated brains.

How it relates to other terms

See also Backpropagation The algorithm that works out how much each weight in a network contributed to the error, by applying the chain... See also Gradient Descent The optimisation method that trains most models: repeatedly step every parameter a little way downhill on the... Narrower term Convolutional Neural Network A network that slides small learned filters across an image, building up from edges to shapes to objects.
Referenced by
Backpropagation Includes Convolutional Neural Network Narrower term Recurrent Neural Network Narrower term

Related terms

Overfitting

When a model learns the training data so closely — including its noise — that it performs worse on anything new.

Beginner Machine Learning

Embedding

A list of numbers representing a piece of data, arranged so that similar things end up close together.

Intermediate Machine Learning

Was this definition helpful?

Definitions are reviewed by our editorial team. Spotted a problem? Tell us.

Keep reading

Browse the glossary Suggest a correction