Online Alternating Direction Method of Multipliers for Online Composite Optimization

MaRDI QIDQ6503436arXivFDO

Authors Yu-Le Zhang, Zehao Xiao, Jia Wu, Liwei Zhang

Published in arXiv

Abstract: In this paper, we investigate regrets of an online semi-proximal alternating direction method of multiplier (Online-spADMM) for solving online linearly constrained convex composite optimization problems. Under mild conditions, we establish $mO(sqrtN)$ objective regret and $mO(sqrtN)$ constraint violation regret at round $N$ when the dual step-length is taken in $(0,(1+sqrt5)/2)$ and penalty parameter $sigma$ is taken as $sqrtN$. We explain that the optimal value of parameter $sigma$ is of order $mO(sqrtN)$. Like the semi-proximal alternating direction method of multiplier (spADMM), Online-spADMM has the advantage to resolve the potentially non-solvability issue of the subproblems efficiently. We show the usefulness of the obtained results when applied to online quadratic optimization problem. The inequalities established for Online-spADMM are also used to develop iteration complexity of the average update of spADMM for solving linearly constrained convex composite optimization problems.

This page was built for publication: Online Alternating Direction Method of Multipliers for Online Composite Optimization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6503436)