The neuron, the layer, the network

A neuron is a weighted sum of its inputs, passed through a non-linear squashing function.

Neuron

output = f(w₁ x₁ + w₂ x₂ + … + b)

Layer

a row of neurons, each looking at the same inputs

Network

stacked layers; output of one feeds the next

Training

nudge the weights so predictions improve

input

hidden layers

output