Rida Laraki – Mathematics and Decision

$\begin{quote} \begin{center} \textbf{Online Learning in Identical-Interest and Zero-Sum Discounted Stochastic Games} \end{center} \medskip This presentation combines two recent articles where we extend a popular decentralized discrete-time learning procedure called fictitious play (Brown, 1951; Robinson, 1951) to discounted stochastic games (Shapley, 1953). We prove the convergence of our online learning algorithms to the set of stationary Nash equilibria in identical-interest and zero-sum discounted stochastic games. The two articles in question are: \smallskip \begin{itemize} \item[1] Lucas Baudin and Rida Laraki, ``Fictitious Play and Best-Response Dynamics in Identical Interest and Zero-Sum Stochastic Games", \textit{Proceedings of the 39th International Conference on Machine Learning}, PMLR 162:1664-1690, 2022. \smallskip \item[2] Lucas Baudin and Rida Laraki. ``Smooth Fictitious Play in Stochastic Games with Perturbed Payoffs and Unknown Transitions." \textit{Advances in Neural Information Processing Systems} 35 (2022): 20243-20256. \end{itemize} \end{quote}$