Almost Sure Saddle Avoidance of Stochastic Gradient Methods without the Bounded Gradient Assumption

Devs

Almost Sure Saddle Avoidance of Stochastic Gradient Methods without the Bounded Gradient Assumption | Read Paper on Bytez