NeurIPS 2020spotlight18 citations

Tight First- and Second-Order Regret Bounds for Adversarial Linear Bandits

Shinji Ito, Shuichi Hirahara, Tasuku Soma, Yuichi Yoshida

Abstract

We propose novel algorithms with first- and second-order regret bounds for adversarial linear bandits. These regret bounds imply that our algorithms perform well when there is an action achieving a small cumulative loss or the loss has a small variance. In addition, we need only assumptions weaker than those of existing algorithms; our algorithms work on discrete action sets as well as continuous ones without a priori knowledge about losses, and they run efficiently if a linear optimization oracle for the action set is available. These results are obtained by combining optimistic online optimization, continuous multiplicative weight update methods, and a novel technique that we refer to as distribution truncation. We also show that the regret bounds of our algorithms are tight up to polylogarithmic factors.

BibTeX
@inproceedings{NEURIPS2020_15bb63b2,
 author = {Ito, Shinji and Hirahara, Shuichi and Soma, Tasuku and Yoshida, Yuichi},
 booktitle = {Advances in Neural Information Processing Systems},
 editor = {H. Larochelle and M. Ranzato and R. Hadsell and M.F. Balcan and H. Lin},
 pages = {2028--2038},
 publisher = {Curran Associates, Inc.},
 title = {Tight First- and Second-Order Regret Bounds for Adversarial Linear Bandits},
 url = {https://proceedings.neurips.cc/paper_files/paper/2020/file/15bb63b28926cd083b15e3b97567bbea-Paper.pdf},
 volume = {33},
 year = {2020}
}