Backpropagation
Domain: Math & Logic for AI
🎯 Motivation
The algorithm that computes gradients for all weights by applying the chain rule backward through the network.
📋 Concrete Example
Loss is computed at the output. Backprop sends the error backward, computing each layer contribution using the chain rule.
🔗 Analogy
Like a factory quality control running backward through the line, blaming each station for its share of defects.