Groups
Category
Weight initialization sets the starting values of neural network parameters so signals and gradients neither explode nor vanish as they pass through layers.
Probability theory formalizes uncertainty using a sample space, events, and a probability measure that obeys clear axioms.