NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:3818
Title:Beyond Confidence Regions: Tight Bayesian Ambiguity Sets for Robust MDPs

The paper proposes a novel approach for robust MDPs. The reviewers found the paper interesting, but difficult to read. The theoretical contribution is significant and the empirical contribution is adequate to support the theory. If the paper is accepted, the authors are strongly encouraged to follow the advice of the reviewers to improve the readability of the paper.