S ECANT: Self-Expert Cloning for Zero-Shot Generalization of Visual Policies

               Linxi Fan 1 2 * Guanzhi Wang 1 De-An Huang 2 Zhiding Yu 2
                  Li Fei-Fei 1 Yuke Zhu 3 2 Anima Anandkumar 4 2

             Abstract
   Generalization has been a long-standing challenge
   for reinforcement learning (RL). Visual RL, in
   particular, can be easily distracted by irrelevant
   factors in high-dimensional observation space.
   In this work, we consider robust policy learning
   ...