ConceptAI & ML
Multiclass SVM loss initialization
When initializing weights to small random values, scores are approximately zero and the initial loss equals the number of classes minus one.
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.
Lecturers
1Connections
2The lectures that define it or assume it, and the concepts next to it.