1. events
  2. all
  3. 2022

Jul 26 (Tue) @ 9:00am: "Strategic Mechanisms in Multi-Agent Coordination," Rahul Chandan, ECE PhD Defense

Date and Time
Location
Zoom Meeting –

https://ucsb.zoom.us/my/rchandan

Abstract

Strategic interactions in multi-agent systems can be conveniently modeled, manipulated and characterized within the analytical framework provided by game theory and mechanism design. Accordingly, we adopt this framework and pursue research directions related to how information and externalities impact the strategic outcomes that can emerge in systems with multiple, non-cooperative decision makers. In this talk, we will consider two levels of interaction corresponding with the agent- and system-level, and propose variations on the well-studied Colonel Blotto and congestion game models, respectively, to conduct our analysis. In our work on Colonel Blotto games, we quantify the value of information on the system state, and also show that -- perhaps surprisingly -- revealing information to competitors can also offer strategic benefits in competitive resource allocation settings. In our work on congestion games, we focus on the design of taxes to optimize the worst-case equilibrium efficiency (i.e., Price of Anarchy). We derive a tractable linear programming-based approach for computing the optimal taxes within this setting, but show that optimizing for the worst case has significant negative consequences on other relevant performance measures (i.e., best-case equilibrium efficiency, transient behaviour).

Bio

Rahul is a Ph.D. candidate and U.C. Regents' Fellow in the Department of Electrical and Computer Engineering at the University of California, Santa Barbara, working under the supervision of Dr. Jason Marden. He previously received his M.S. in the Department of Electrical and Computer Engineering at the University of California, Santa Barbara in 2019, and his B.A.Sc. in Engineering Science at the University of Toronto in 2017. Rahul's research interests are in the applications of game theory, learning and optimization to problems in engineering, computer science and operations research.

Hosted by: Professor Jason R. Marden

Submitted by: Rahul Chandan <rchandan@ucsb.edu>