An optimal algorithm for stochastic strongly-convex optimization

摘要

We consider stochastic convex optimization with a strongly convex (but notnecessarily smooth) objective. We give an algorithm which performs onlygradient updates with optimal rate of convergence.