Distributional Robustness and Regularization in Reinforcement Learning

Esther Derman, Shie Mannor

Distributionally Robust Optimization (DRO) has enabled to prove the equivalence between robustness and regularization in classification and regression, thus providing an analytical…