Tighter Problem-Dependent Regret Bounds in Reinforcement Learning without Domain Knowledge using Value Function Bounds.

Andrea Zanette, Emma Brunskill

Browse the full ICML paper archive.