Backpropagation Basics

Backpropagation is how neural networks learn by sending error signals backward to update weights via the chain rule.