Core

Neural Network

A computational model composed of layers of interconnected nodes (neurons) that process data through weighted connections, learning to approximate complex functions.

Explained at five levels

Level 1

A computer brain made of tiny connected pieces that work together to figure things out, kind of like how your brain works.

Level 2

A computer system inspired by the human brain — lots of small parts called "neurons" connected together that learn to recognize patterns.

Level 3

A computational model composed of layers of interconnected nodes (neurons) that process data through weighted connections, learning to approximate complex functions.

Level 4

A parameterized function approximator organized in layers of artificial neurons with learnable weights, trained via backpropagation and gradient descent to minimize a loss function.

Level 5

A directed computational graph of differentiable operators where each node applies an affine transformation followed by a nonlinear activation, trained end-to-end via stochastic gradient descent on a loss surface with complex geometry.

Definitions are educational summaries. Terminology can vary by source and context.

Sources