Deep Reinforcement Learning (DRL) has achieved expressive success across a wide range
of domains. However, it is still faced with the sample-inefficiency problem that requires …