The Mirage of Action-Dependent Baselines in Reinforcement Learning.

George Tucker, Surya Bhupatiraju, Shixiang Gu, Richard E. Turner, Zoubin Ghahramani, Sergey Levine

Browse the full ICML paper archive.