报告题目:基于结构化通信的多智能体学习及应用
Multi-agent Learning and Applications based on Structured Communication
报告人:王祥丰 副教授 (华东师范大学计算机科学与技术学院)
报告时间:2022年10月28日(周五)上午9:00-10:00
报告地点:腾讯会议
会议ID :105-620-973
校内报告联系人:刘永朝 教授 联系电话:84708351-8141
报告摘要:This work explores the large-scale multi-agent communication mechanism for multi-agent reinforcement learning (MARL). We summarize the general categories of topology for communication structures, which are often manually specified in MARL literature. A novel framework termed Learning Structured Communication (LSC) is proposed by utilizing flexible and efficient communication topology. Our framework allows for adaptive agent grouping to form different hierarchical formations over episodes, which is generated by an auxiliary task combined with a hierarchical routing protocol. Given the formed topology, a hierarchical graph neural network is learned to enable effective message information generation and propagation among inter- and intra-group communications. In contrast to existing communication mechanisms, our method has an explicit while the learnable design for hierarchical communication. Experiments on challenging tasks, e.g., multi-agent path-finding of AGVs, show the proposed LSC performs high communication efficiency and global cooperation capability.
报告人简介:王祥丰,华东师范大学计算机科学与技术学院副教授,2009年和2014年分别获得南京大学学士和博士学位;攻读博士学位期间,获得国家留学基金委资助赴美国明尼苏达大学联合培养。毕业后,加入华东师范大学计算机科学与技术学院,主要研究方向是多智能体学习、分布式优化、可信机器学习等。获得2021年IEEE信号处理学会最佳论文奖,2022年入选上海市青年科技英才启明星。已在IEEE TPAMI、IEEE T-Cybernetics、IEEE TSP、IEEE TMI、PR等人工智能国际权威期刊以及ICLR、SIGKDD、CVPR、IJCAI、ICRA、AAMAS、UAI、ICMR等人工智能国际权威会议发表论文30余篇;并在Mathematical Programming、Mathematics of Operations Research、SIAM Journal on Scientific Computing等运筹学国际权威期刊发表论文。