Sample-Efficient and Safe Deep Reinforcement Learning via Reset Deep Ensemble Agents

We show that the proposed method significantly improves the performance of the baseline algorithm in the considered environments.