主 讲 人：魏晓利  哈尔滨工业大学 副教授

报告时间：2025年12月8日 上午10:00-11:00

报告地点：腾讯会议 803-340-191

 

报告摘要:  In this talk, we present a series of results on reinforcement learning (RL) for mean-field control (MFC) problems. We begin with the study of discrete-time Q-learning algorithms for MFC. We then investigate q-learning, the recently coined as the continuous time counterpart of Q-learning by Jia and Zhou (2023) in the framework of entropy-regularized reinforcement learning. Our work addresses three fundamental questions in RL for MFC problems: (1) how to define an appropriate form of the Q-function (or q-function), (2) how to derive optimal policies from the Q-function (or q-function), and (3) how to learn the Q-function (or q-function) itself.

 

主讲人简介：魏晓利，哈尔滨工业大学准聘副教授。本科毕业于中国科学技术大学，2018年于巴黎第七大学获得博士学位。2019-2021年在加州大学伯克利分校从事博士后。2021年-2023年在清华大学深圳国际研究生院担任助理教授。主要从事随机微分博弈、强化学习等研究。论文发表在Operations Research，Mathematical Finance, SIAM Journal on Control and Optimization等期刊杂志。