Litcius/Paper detail

Deep Reinforcement Learning based Video Games: A Review

Kamal A. ElDahshan, Hesham Farouk, Eslam Mofreh

20222022 2nd International Mobile, Intelligent, and Ubiquitous Computing Conference (MIUCC)15 citationsDOI

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.

Topics & Concepts

Reinforcement learningComputer scienceArtificial intelligenceVideo gameVariety (cybernetics)Deep learningSet (abstract data type)ReinforcementMultimediaEngineeringStructural engineeringProgramming languageReinforcement Learning in RoboticsArtificial Intelligence in GamesAdvanced Bandit Algorithms Research