Backpropagation
The algorithm that computes how much each weight in a network contributed to the final error, by applying the chain rule backward from the output layer to the input layer. Those per-weight gradients are what gradient descent then uses to update the weights.
Why exams ask this
Tested conceptually, not as a derivation. The exam wants "the chain rule applied backward through the network to compute each weight's gradient," and flags an answer that describes it as simply "the model learning from its mistakes" with no mechanism named.
Relevant to
Related concepts
Resources
No resources linked to this concept yet.