Shie Mannor

Theory of Fair Representations, and a computationally tractable fairness performance Pareto front algorithm.

We imrove sample complexity of single time scale actor critic to $O(\epsilon^{-3})$ from $O(\epsilon^{-4})$ for obtaining $\epsilon$-close global optimal policy.

We propose a efficient robust policy evaluation method for non-rectangular robust MDPs with uncertainty sets bounded by $L_p$ norms.