6.2.3. Neural Networks

Neural networks are a family of models loosely inspired by the brain: individual computational units (neurons) are connected into layers, and the network learns by adjusting the strength (weights) of those connections. Starting from a single neuron, we can stack layers of thousands of neurons and train models of remarkable expressive power.

This chapter builds from first principles, starting with the simplest possible “network” - a single perceptron - and progressively adding depth and complexity.