机器学习与数据科学博士生系列论坛（第六十三期）—— Information Design in Games: Stackelberg Equilibrium
报告人：陈雨静（北京大学）
时间：2023-12-07 16:00-17:00
地点：腾讯会议 551-1675-5419
摘要：
Many problems in economic theory involve sequential reasoning between multiple parties with asymmetric access to information. This broad class of problems leads to many research questions about information design and optimal strategic behaviors. At a high level, information design can be seen as a Stackelberg game. Thus, we can consider information design from a game-theoretic perspective.
In this talk, we first discuss the relation between information design and the Stackelberg game, also considering the case in multiplayer. In the second part, we focus on learning Stackelberg equilibria in games. The goal is to approximate the leader's optimal strategy. Recent research consists of robust Stackelberg equilibria and sample-efficient learning in reinforcement learning.
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。
