REGAL: A Regularization based Algorithm for Reinforcement Learning in Weakly Communicating MDPs.

Peter L. Bartlett, Ambuj Tewari

Browse the full UAI paper archive.