AI glossary
What is backpropagation?
Short answer
The algorithm that computes every weight's gradient by passing error signals backwards through a network using the chain rule.
For example
If a network says "dog" for a photo of a cat, backpropagation works out how much each weight contributed to that mistake, so every weight can be nudged a little in the right direction.
Learn it in a free lesson
Related terms
Questions
What is backpropagation in simple terms?
The algorithm that computes every weight's gradient by passing error signals backwards through a network using the chain rule.
What is an example of backpropagation?
If a network says "dog" for a photo of a cat, backpropagation works out how much each weight contributed to that mistake, so every weight can be nudged a little in the right direction.
Where can I learn about backpropagation for free?
AI in Bed explains it in the free lesson "Backpropagation: blame, passed backwards" (Intermediate level, 6 minutes), with no sign-up.