Convergence of Stochastic Gradient Method for a Class of Nonsmooth Nonconvex Optimization

In this paper, we study a (sub)-gradient-based method for solving a class of stochastic optimization problems where the objective consists of a difference of smooth function (possibly nonconvex) and a convex function. This general class contains many stochastic optimization problems. We establish novel convergence result for stochastic (sub)- gradient descent for solving this class of stochastic problems

کلیدواژه ها:

Stochastic Gradient ، Nonsmooth Nonconvex