Glossary · Model Training

Backpropagation

The algorithm that computes how much each parameter in a neural network contributed to its prediction error, enabling gradient descent to update parameters efficiently across many layers.