报告人：Jingxin Zhan (PKU)

时间：2023-11-09 16:00-17:00

地点：腾讯会议 551-1675-5419

摘要： Adaptive bandit algorithms that automatically exploit certain specific characteristics of environments have been studied for many years. At the higher level, the proposed algorithm adapts to a variety of types of environments, i.e., adversarial environments and stochastic environments. At the lower level, in each of the adversarial and stochastic regimes, the proposed algorithm adapts to certain environmental characteristics, for example, the total quadratic variation in loss sequence, thereby performing better.

In this talk, we will introduce adaptive bandit problems in the linear bandit setting, following the recent two works from COLT. They both apply the framework of follow-the-regularized-leader (FTRL) methods, which has already yielded many important results in the original bandit setting.

论坛简介：该线上论坛是由张志华教授机器学习实验室组织，每两周主办一次（除了公共假期）。论坛每次邀请一位博士生就某个前沿课题做较为系统深入的介绍，主题包括但不限于机器学习、高维统计学、运筹优化和理论计算机科学。