A Comprehensive Discussion on Deep Reinforcement Learning

Deep reinforcement learning (RL) is an important method to provide solutions for the perception and decision-making problems of complex systems. It has provided great contributions in natural language processing, drug design, gaming, and some other fields. This paper gives an overview of recent achievements in deep reinforcement learning. First, some basic concepts regarding deep learning, reinforcement learning and deep reinforcement learning are first introduced in this paper. Then, the milestone related work of deep RL and two methods of deep RL: deep Q-learning-based and policy gradient-based are reviewed. At last, an overview of deep RL applications is presented, and the future of deep RL is also proposed.