Quantile Markov Decision Process (1711.05788v4)

Abstract: The goal of a traditional Markov decision process (MDP) is to maximize expected cumulativereward over a defined horizon (possibly infinite). In many applications, however, a decision maker may beinterested in optimizing a specific quantile of the cumulative reward instead of its expectation. In this paperwe consider the problem of optimizing the quantiles of the cumulative rewards of a Markov decision process(MDP), which we refer to as a quantile Markov decision process (QMDP). We provide analytical resultscharacterizing the optimal QMDP value function and present a dynamic programming-based algorithm tosolve for the optimal policy. The algorithm also extends to the MDP problem with a conditional value-at-risk(CVaR) objective. We illustrate the practical relevance of our model by evaluating it on an HIV treatmentinitiation problem, where patients aim to balance the potential benefits and risks of the treatment.