Abstract: Nash equilibria are sought-after outcomes in a non-cooperative multi-agent system as no agent finds it profitable to unilaterally deviate from such an outcome. How can players learn equilibrium strategies based only on evaluation of their objective functions, rather than closed-form expressions of them? What can players learn in the case in which the game does not admit a Nash equilibrium? Such questions arise in applications ranging from auctions, network routing to power markets. I will discuss our approaches to address these problems in static convex and non-convex games. Leveraging these results, I will discuss our recent work on learning equilibria in multi-agent reinforcement learning problems.
Bio: Maryam Kamgarpour holds a Doctor of Philosophy in Engineering from the University of California, Berkeley and a Bachelor of Applied Science from University of Waterloo, Canada. Her research is on safe decision-making and control under uncertainty, game theory and mechanism design, mixed integer and stochastic optimization and control. Her theoretical research is motivated by control challenges arising in intelligent transportation networks, robotics, power grid systems and healthcare. She is the recipient of NASA High Potential Individual Award, NASA Excellence in Publication Award, and the European Union (ERC) Starting Grant.