English

Distributionally Robust Counterpart in Markov Decision Processes

Systems and Control 2015-05-14 v2 Optimization and Control

Abstract

This paper studies Markov Decision Processes under parameter uncertainty. We adapt the distributionally robust optimization framework, and assume that the uncertain parameters are random variables following an unknown distribution, and seeks the strategy which maximizes the expected performance under the most adversarial distribution. In particular, we generalize previous study \cite{xu2012distributionally} which concentrates on distribution sets with very special structure to much more generic class of distribution sets, and show that the optimal strategy can be obtained efficiently under mild technical condition. This significantly extends the applicability of distributionally robust MDP to incorporate probabilistic information of uncertainty in a more flexible way.

Keywords

Cite

@article{arxiv.1501.07418,
  title  = {Distributionally Robust Counterpart in Markov Decision Processes},
  author = {Pengqian Yu and Huan Xu},
  journal= {arXiv preprint arXiv:1501.07418},
  year   = {2015}
}

Comments

Added references. Corrected typos. Modified a mistake in Example 2 (Variance). Provided more details of the simulation

R2 v1 2026-06-22T08:15:41.211Z