Spotlight
Online Markov Decision Processes under Bandit Feedback
Gergely Neu · András György · András Antos · Csaba Szepesvari

Wed Dec 8th 11:25 -- 11:30 AM @ Regency Ballroom
in Spotlights Session 9 »

We consider online learning in finite stochastic Markovian environments where in each time step a new reward function is chosen by an oblivious adversary. The goal of the learning agent is to compete with the best stationary policy in terms of the total reward received. In each time step the agent observes the current state and the reward associated with the last transition, however, the agent does not observe the rewards associated with other state-action pairs. The agent is assumed to know the transition probabilities. The state of the art result for this setting is a no-regret algorithm. In this paper we propose a new learning algorithm and assuming that stationary policies mix uniformly fast, we show that after T time steps, the expected regret of the new algorithm is O(T^{2/3} (ln T)^{1/3}), giving the first rigorously proved convergence rate result for the problem.

Author Information

Gergely Neu (Universitat Pompeu Fabra)
András György (University of Alberta)
András Antos (MTA SZTAKI Institute for Computer Science and Control)
Csaba Szepesvari (DeepMind / University of Alberta)

More from the Same Authors