Reward-Mixing MDPs with Few Latent Contexts are Learnable.

Jeongyeol Kwon, Yonathan Efroni, Constantine Caramanis, Shie Mannor

Browse the full ICML paper archive.