Online Optimization with Memory and Competitive Control
Machine Learning
2021-01-11 v3 Systems and Control
Systems and Control
Optimization and Control
Machine Learning
Abstract
This paper presents competitive algorithms for a novel class of online optimization problems with memory. We consider a setting where the learner seeks to minimize the sum of a hitting cost and a switching cost that depends on the previous decisions. This setting generalizes Smoothed Online Convex Optimization. The proposed approach, Optimistic Regularized Online Balanced Descent, achieves a constant, dimension-free competitive ratio. Further, we show a connection between online optimization with memory and online control with adversarial disturbances. This connection, in turn, leads to a new constant-competitive policy for a rich class of online control problems.
Cite
@article{arxiv.2002.05318,
title = {Online Optimization with Memory and Competitive Control},
author = {Guanya Shi and Yiheng Lin and Soon-Jo Chung and Yisong Yue and Adam Wierman},
journal= {arXiv preprint arXiv:2002.05318},
year = {2021}
}
Comments
Neural Information Processing Systems (NeurIPS 2020)