Optimistic Regret Bounds for Online Learning in Adversarial Markov Decision Processes
Sang Bin Moon, Abolfazl Hashemi
TL;DR
This work addresses AMDPs where time-varying costs may not be truly adversarial by integrating a set of cost predictors into policy learning. It introduces a novel optimistically biased, variance-reducing cost estimator and an OREPS-OPIX policy-search update that achieves high-probability optimistic regret bounds, with tighter guarantees when cost predictors are accurate. The main results provide full-information and bandit-regret bounds that degrade gracefully with predictor error, plus an anytime extension and a treatment of unknown transitions via confidence sets. Numerical experiments on a drone-navigation AMDP demonstrate reduced regret and variance with the proposed estimator, highlighting practical benefits for non-stationary, adversarial-like environments. Overall, the approach offers a principled way to leverage predictive signals to obtain favorable regret guarantees in AMDPs across feedback regimes and settings.
Abstract
The Adversarial Markov Decision Process (AMDP) is a learning framework that deals with unknown and varying tasks in decision-making applications like robotics and recommendation systems. A major limitation of the AMDP formalism, however, is pessimistic regret analysis results in the sense that although the cost function can change from one episode to the next, the evolution in many settings is not adversarial. To address this, we introduce and study a new variant of AMDP, which aims to minimize regret while utilizing a set of cost predictors. For this setting, we develop a new policy search method that achieves a sublinear optimistic regret with high probability, that is a regret bound which gracefully degrades with the estimation power of the cost predictors. Establishing such optimistic regret bounds is nontrivial given that (i) as we demonstrate, the existing importance-weighted cost estimators cannot establish optimistic bounds, and (ii) the feedback model of AMDP is different (and more realistic) than the existing optimistic online learning works. Our result, in particular, hinges upon developing a novel optimistically biased cost estimator that leverages cost predictors and enables a high-probability regret analysis without imposing restrictive assumptions. We further discuss practical extensions of the proposed scheme and demonstrate its efficacy numerically.