T Dick, A György, C Szepesvári - … of the 31st International Conference on …, 2014 - dl.acm.org
In this paper we consider online learning in finite Markov decision processes (MDPs) with
changing cost sequences under full and bandit-information. We propose to view this …