Decision and Control Fall Seminar Series: Vijay Subramanian || Games in Multi-Agent Dynamic Systems: Decision-Making with Compressed Information

Abstract: The model of multi-agent dynamic systems has a wide range of applications in numerous socioeconomic and engineering settings: spectrum markets, e-commerce, transportation networks, power systems, etc. In this model, each agent takes actions over time to interact with the underlying system as well as each other to achieve their respective objectives. In many applications of this model, agents have access to a huge amount of information that increases over time. Determining solutions of such multi-agent dynamic games can be complicated due to the huge domains of strategies as well as the intertwining of beliefs and strategies. Meanwhile, agents have restrictions on their computational power and communication capability as well as latency limitations, which prevent them from implementing complicated strategies. Therefore, it is important to identify suitable compression schemes so that at equilibrium each agent can make decisions based on a compressed version of their information instead of the full information. However, compression of information could result in loss of some or all equilibrium outcomes. This talk presents results on this issue for a general class of multi-agent dynamic games, and designs and analyzes appropriate information compression schemes. Our results highlight the tension among information compression, preservation of  equilibrium outcomes, and applicability of sequential decomposition algorithms to find compression-based equilibria. This is joint work with Dengwang Tang (University of California, Berkeley) and Demos Teneketzis (University of Michigan).

Bio: Vijay Subramanian received the Ph.D. degree in electrical engineering from the University of Illinois at Urbana-Champaign,  Champaign, IL, USA, in 1999. He worked at Motorola Inc., at the Hamilton Institute, Maynooth, Ireland, for many years, and also in the EECS Department, Northwestern University, Evanston, IL, USA. In Fall 2014, he started in his current position as an Associate Professor with the  EECS Department at the University of Michigan, Ann Arbor. For the academic year 2022-2023, he’s visiting UIUC hosted graciously by CSL and also the ECE department. His research interests are in stochastic analysis, random  graphs, multi-agent systems, and game theory (mechanism and information design) with applications to social, economic and technological networks.