Policy Gradient with Tree Search (PGTS) in Reinforcement Learning Evades Local Maxima.

Navdeep Kumar, Priyank Agrawal, Kfir Yehuda Levy, Shie Mannor

Browse the full ICLR paper archive.