Deep Reinforcement Learning based Video Games: A Review
Kamal A. ElDahshan, Hesham Farouk, Eslam Mofreh
Abstract
Video game development is getting increasingly effective as AI paradigms advance. Deep Reinforcement Learning (DRL) is a promising artificial intelligence (AI) approach. It broadens the reinforcement learning paradigm by making use of the complicated representation of deep neural networks. According to recent research, DRL performs extraordinarily well across a variety of sectors, including healthcare, video games, robotics, finance, and medicine. In this article, we provide a complete review of current and cutting-edge research developments in deep reinforcement learning in video games. This review starts by discussing the principles of reinforcement and deep reinforcement learning. Then investigate and assess essential strategies such as value-based, policy gradient, and model-based algorithms in this work. Finally, discusses a set of research challenges of DRL in video games.