Learning and Solving Games in the Presence of Teams
September 24, 2024 3:30 PM Singapore (Registration starts at 3:20 PM)
Abstract
One challenge in multiagent environments is the presence of teams — groups of agents acting in a coordinated fashion that distinguish themselves from individuals. This talk touches on two situations where teams are prevalent. The first is multi-defender security games, where heterogeneous players distribute defensive resources in a manner that is incentive compatible, while accounting for a best-responding follower (e.g., military alliances and potential adversaries). We describe sufficient conditions for equilibrium to exist and describe polynomial time solvers. The second situation is Coalition Structure Learning (CSL), where one has to learn the underlying coalition structure in many-agent environments by actively designing “query games” and observing the resultant equilibrium. We derive lower bounds on the number of queries and present algorithms matching these bounds.
PAPERS:
https://arxiv.org/abs/2311.16392
https://arxiv.org/abs/2312.09058
About the Speaker
Chun Kai Ling is an Assistant Professor in the Department of Computer Science in NUS. His research is on multiagent systems and computational game theory. He develops methods in machine learning and optimization to tackle real-world problems such as security and logistics. He currently works on modeling and solving large games, featuring multiple players or general-sum utilities. His work on inverse game theory and multi-defender security games were best papers in IJCAI 2018 and GameSec 2023. He was awarded the Sung Kah Kay Assistant Professorship in 2024 and NUS development grant from 2022-2024. Prior to joining NUS, he was a Postdoc at Columbia University working with Christian Kroer and Garud Iyengar. He completed his PhD at Carnegie Mellon University under the supervision of Zico Kolter and Fei Fang.
For more information about the ESD Seminar, please email esd_invite@sutd.edu.sg