Computer Games: 7th Workshop, CGW 2018 Held in Conjunction with the 27th International Conference on Artificial Intelligence, IJCAI 2018 Stockholm, Sweden, July 13, 2018: Revised Selected Papers

Description

This book constitutes revised selected papers from the 7th Workshop on Computer Games, CGW 2018, held in conjunction with the 27th International Conference on Artificial Intelligence, IJCAI 2018 in Stockholm, Sweden, in July 2018. The 8 full papers presented in this volume were carefully reviewed and selected from 15 submissions. They cover a wide range of topics related to video games; general game playing.- machine learning and Monte Carlo tree search. Preface Organization Contents Video Games Towards Embodied StarCraft II Winner Prediction 1 Introduction 2 Background and Related Work 2.1 StarCraft Series 2.2 Outcome Prediction 2.3 Research Questions 3 Experimental Setup 3.1 Data Collection 3.2 Preprocessing 3.3 Data Characterisation 3.4 Prediction Methods 3.5 Training 4 Experimental Analysis 5 Conclusion and Outlook References MOBA-Slice: A Time Slice Based Evaluation Framework of Relative Advantage Between Teams in MOBA Games 1 Introduction 2 MOBA-Slice 2.1 MOBA Game Result Analysis 2.2 Discounted Evaluation 2.3 Time Slice Evaluation Model 3 Experiments 3.1 Apply MOBA-Slice to DotA2 3.2 Data Processing 3.3 First Training of TSE Model 3.4 Prediction on Continuous Time Slices 3.5 Comparison with DotA Plus Assistant on Result Prediction 3.6 Towards Better Performance 4 Related Work 5 Conclusion References TextWorld: A Learning Environment for Text-Based Games 1 Introduction 2 Text Games from a Reinforcement Learning Perspective 2.1 Text-Based Games 2.2 Text-Based Games as POMDPs 2.3 RL Challenges in Text-Based Games 2.4 RL with TextWorld 3 The TextWorld Learning Environment 3.1 Game Engine 3.2 Game Generation 3.3 Game Interaction with TextWorld 4 Related Work 4.1 Relevant Models 4.2 Frameworks 5 Benchmarks 5.1 Curated List 5.2 Treasure Hunter 6 Current Limitations 7 Conclusion A Typical Text-Based Obstacles B Curated List of Text-Based Games B.1 Game Notes C Inform 7 References General Game Playing Statistical GGP Game Decomposition 1 Introduction 2 Previous Works 3 Game, Subgame and Correct Decomposition 4 Method 4.1 Simulation Based Detection of Action Effects 4.2 Filtering Action Effects 4.3 Action Independent Fluents 4.4 Compound Moves and Meta-action Sets 4.5 Serial Games and crosspoints Identification 4.6 Building the Dependency Graph 4.7 Subgoals to Fix Under- or Over-Decomposition 5 Experiments 6 Conclusion and Future Work References Iterative Tree Search in General Game Playing with Incomplete Information 1 Introduction 2 Background 2.1 General Game Playing with Incomplete Information 2.2 Fictitious Play 3 Iterative Tree Search 3.1 Initialising the Tree 3.2 Iterative Probability Update 4 Analysis 4.1 Games with Dominant Pure Strategy and Single Player Games 4.2 Non-locality Problem 4.3 One-Step Joint-Move Two-Player Zero-Sum Games 4.4 Move Separable Games 5 Conclusion References Machine Learning and Monte Carlo Tree Search Spatial Average Pooling for Computer Go 1 Introduction 2 Training a Value Network 3 PUCT 4 Spatial Average Pooling 5 Experimental Results 6 Conclusion References Analyzing the Impact of Knowledge and Search in Monte Carlo Tree Search in Go 1 Introduction 2 Knowledge and Simple Features 3 Evaluation Methods 3.1 Playing Strength 3.2 Move Prediction 4 Research Motivations 5 Experimental Results and Discussion 5.1 Fuego-Based Players Used in Our Experiments 5.2 Move Prediction 5.3 Playing Strength 5.4 A Closer Look at Move Prediction Rate 5.5 Move Prediction and Feature Frequency 5.6 Move Selection Analysis 6 Conclusions and Future Work A Detailed Move Prediction Results References What's in a Game? The Effect of Game Complexity on Deep Reinforcement Learning 1 Introduction 2 Background 3 Factors that Define a Game 4 Maze Experiments 4.1 The Effect of the Size of the Search Space 4.2 The Effect of Obstacles 4.3 The Effect of Hostile Agents 4.4 The Effect of Intermediate Reward 5 Pacman Experiments 6 Conclusions and Future Work References Author Index