Quantcast

Adaptive Mirror Descent for Constrained Optimization

Research paper by Anastasia Bayandina

Indexed on: 04 May '17Published on: 04 May '17Published in: arXiv - Mathematics - Optimization and Control



Abstract

This paper seeks to address how to solve non-smooth convex and strongly convex optimization problems with functional constraints. The introduced Mirror Descent (MD) method with adaptive stepsizes is shown to have a better convergence rate than MD with fixed stepsizes due to the improved constant. For certain types of constraints, the method is proved to generate dual solution. For the strongly convex case, the restart technique is applied.