Abstract: Deep reinforcement learning (DRL) has shown significant success in domains such as computer vision and robot control. However, DRL agents often suffer from low sample efficiency, limiting ...