Y.-G. Hsieh, K. Antonakopoulos, and P. Mertikopoulos. In COLT '21: Proceedings of the 34th Annual Conference on Learning Theory, 2021.
In game-theoretic learning, several agents are simultaneously following their individual interests, so the environment is non-stationary from each player’s perspective. In this context, the performance of a learning algorithm is often measured by its regret. However, no-regret algorithms are not created equal in terms of game-theoretic guarantees: depending on how they are tuned, some of them may drive the system to an equilibrium, while others could produce cyclic, chaotic, or otherwise divergent trajectories. To account for this, we propose a range of no-regret policies based on optimistic mirror descent, with the following desirable properties: i) they do not require any prior tuning or knowledge of the game; ii) they all achieve $\mathcal{O}(\sqrt{T})$ regret against arbitrary, adversarial opponents; and iii) they converge to the best response against convergent opponents. Also, if employed by all players, then iv) they guarantee $\mathcal{O}(1)$ social regret; while v) the induced sequence of play converges to Nash equilibrium with $\mathcal{O}(1)$ individual regret in all variationally stable games (a class of games that includes all monotone and convex-concave zero-sum games).
arXiv link: https://arxiv.org/abs/2104.12761