Compressive Study on Self-Paced Priority Curriculum Learning by Coverage Penalty in Deep Curriculum Reinforcement Learning

Abstract

In this paper, another preparation worldview is proposed for deep reinforcement learning utilizing self-paced prioritized curriculum learning with coverage penalty. The proposed Deep Curriculum Reinforcement Learning (DCRL) takes the most favorable position of experience replay by adaptively choosing suitable advances from replay memory dependent on the unpredictability of each progress.