RAAM: The Benefits of Robustness in Approximating Aggregated MDPs in Reinforcement Learning

41 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Safe Policy Improvement by Minimizing Robust Baseline Regret (2016, background): “One of the motivations for considering ( x, a ) -rectangular sets in RMDP is that they lead to tractable solutions in the conventional reward maximization setting.”
Robust Policy Optimization with Baseline Guarantees (2015, result): “However, the robust solution can be computed very efﬁciently when the uncertainty set is described, as in our formulation, in terms of an L 1 -norm [17]; similar results also exist for the L 2 norm [9].”
High Confidence Off-Policy Evaluation with Models (2016, unknown): “Model bias in MDPs can come from two sources: 1.”