An InformationTheoretic Approach to Minimax Regret in Partial Monitoring
[edit]
Proceedings of the ThirtySecond Conference on Learning Theory, PMLR 99:21112139, 2019.
Abstract
We prove a new minimax theorem connecting the worstcase Bayesian regret and minimax regret under finiteaction partial monitoring with no assumptions on the space of signals or decisions of the adversary. We then generalise the informationtheoretic tools of Russo and Van Roy (2016) for proving Bayesian regret bounds and combine them with the minimax theorem to derive minimax regret bounds for various partial monitoring settings. The highlight is a clean analysis of ‘easy’ and ‘hard’ finite partial monitoring, with new regret bounds that are independent of arbitrarily large gamedependent constants and eliminate the logarithmic dependence on the horizon for easy games that appeared in earlier work. The power of the generalised machinery is further demonstrated by proving that the minimax regret for $k$armed adversarial bandits is at most $\sqrt{2kn}$, improving on existing results by a factor of 2. Finally, we provide a simple analysis of the cops and robbers game, also improving best known constants.
Related Material


