ConceptAI & ML
Parameter sharing backpropagation
When parameters are shared across multiple components, backpropagation requires summing the gradients coming from all the different branch dimensions.
Know first
Named as a prerequisite by the lectures that define this. The number is how many say so. A grey one has no page yet.
Branching gradients1Parameter sharing1
Lecturers
1Connections
2The lectures that define it or assume it, and the concepts next to it.