Stabilization of stochastic approximation by step size adaptation

2010/07/27 by Sameer Kamal, Kamal, Sameer
Mathematics · #60 #FOS: Mathematics #Probability (math.PR) #math.PR #msc:60

paper · pdf · doi:10.48550/arxiv.1007.4689

15 pages

arxiv created 2010/07/27 · arxiv updated 2010/07/28

Abstract

A scheme for stabilizing stochastic approximation iterates by adaptively scaling the step sizes is proposed and analyzed. This scheme leads to the same limiting differential equation as the original scheme and therefore has the same limiting behavior, while avoiding the difficulties associated with projection schemes. The proof technique requires only that the limiting o.d.e. descend a certain Lyapunov function outside an arbitrarily large bounded set.

Related

• On the convergence of adaptive approximations for stochastic differential equations
• A stochastic approximation algorithm with multiplicative step size adaptation
• Adaptive step-size control for global approximation of SDEs driven by countably dimensional Wiener process
• Step-Size Stability in Stochastic Optimization: A Theoretical Perspective
• Stationary Behavior of Constant Stepsize SGD Type Algorithms: An Asymptotic Characterization