Tag
The paper proposes dynamic context scheduling to enhance generalization in contextual reinforcement learning, showing improved performance in out-of-distribution and in-distribution regimes across simulation environments.