The Network

Neural Networks

Neural Networks

What is a neural network?

A neural network is made of many simple computing units connected together. Each unit takes in information, performs a small calculation, and passes the result to other units.

By connecting many of these units, a network can learn to recognize patterns in things like images, sounds, and language.

The idea was inspired by how neurons in the brain connect and pass information. But a neural network is a mathematical model, not a miniature brain.

What Is a Neural Network?

A neural network is a collection of connected computational units whose adjustable parameters — weights — transform inputs into useful outputs. That is the core idea behind how neural networks learn.

What are neural network weights?

Weights control how strongly one unit influences another. During learning, the network strengthens some connections and weakens others so signals flowing through it produce better predictions.

What do layers do?

Layers stack simple units. Early structure can respond to simpler patterns; deeper combinations help the network recognize more complex shapes and relationships than a single unit can.

How neural networks recognize patterns

Pattern recognition emerges from many weighted connections adjusting together — not from a single hand-written rule for every case. You can follow the idea without calculus; the interactive lesson shows the adjustments in action.