Backpropagation
In plain terms
The algorithm that trains neural networks by propagating errors.
Going deeper
Backpropagation computes how each weight contributed to the error and adjusts them via gradients, layer by layer backward. It's how networks learn.
Under the hood
Uses the chain rule to compute gradients efficiently; paired with gradient descent to update weights.
Examples
Neural network training, gradient computation, deep learning, weight updates.
Think of it like...
Like tracing a mistake back through a chain of decisions to correct each contributor.