I am a strong believer in the benefits of reproducible research. Therefore, all the material needed to regenerate my publications and presentations is available on GitHub.
The goal of my research is to enable a group of agents to control a dynamical system in a decentralized fashion through game-theoretic learning. This requires crafting a game whose equilibria are desirable controllers for the dynamical system and equipping the agents with learning rules that guarantee convergence to an equilibrium.
I study stochastic games in which the full-rationality requirement of game theory is restrictive. To circumvent this restriction, I introduced a bounded-rationality solution concept called empirical-evidence equilibrium (EEE). In the EEE framework, agents use low-order empirical models of observed quantities. These models are statistically consistent when the empirical evidence observed by an agent does not contradict its model. In an EEE, each agent has a consistent model of its opponents and plays a best response to its model. I am currently analyzing properties of EEEs.