Information Design and Learning in Strategic Multi-Agent Decision-Making
Abstract: In multi-agent systems, the design of information mechanisms and learning dynamics plays a critical role in shaping long-term outcomes. This talk explores two perspectives on strategic decision-making: how a central entity can design information to influence agent behavior and how decentralized agents learn and adapt in dynamic environments. In the first part, I will discuss a platform-mediated market where agents strategically reposition based on information about market conditions. The platform must design an optimal information disclosure policy to guide resource movement and maximize commission revenues. We show that a simple monotone partitional policy—revealing extreme market states while pooling intermediate ones—is optimal in many practical settings. We also develop algorithmic methods to compute near-optimal disclosure strategies in general settings. The second part of the talk shifts to decentralized learning in general-sum Markov games, where agents update strategies asynchronously and independently. Unlike well-studied zero-sum and potential games, real-world interactions often lack strong structural properties that guarantee convergence. We introduce a decentralized actor-critic learning dynamic and a new analytical framework for characterizing the convergence region in general Markov games.
Biography: Manxi Wu is an assistant professor in the Department of Civil and Environmental Engineering at University of California, Berkeley. Her research interests include game theory, multi-agent learning, and market design with applications in societal scale systems. Previously, she was an assistant professor in the school of Operations Research and Information Engineering at Cornell University. Manxi received the B.S. degree in applied mathematics from Peking University in 2015, and the Ph.D. degree in Social and Engineering Systems from MIT in 2021. Manxi was a research fellow at the Simons Program on Learning and Games, and a postdoctoral scholar at EECS, University of California, Berkeley.