Pattern Recognition in Computer Games Chumphol Bunkhumpornpat, Ph.D.

24 Slides918.85 KB

Pattern Recognition in Computer Games Chumphol Bunkhumpornpat, Ph.D. Department of Computer Science Faculty of Science Chiang Mai University

204453: Pattern Recognition 2

Google AI beats top human players at strategy game StarCraft II AlphaStar was built by Google’s AI firm DeepMind. a machine capable of learning or understanding any task that humans can AlphaStar has 1,026 actions to choose from at any moment. DeepMind limited the speed of AlphaStar’s reflexes to that of experienced human players. AlphaStar placed within the top 0.5% of all players on the European server. 204453: Pattern Recognition 3

StarCraft II players battle each other in a futuristic warzone. 204453: Pattern Recognition 4

Introduction Computer games are an unique application area for pattern recognition. Challenging synthetic opponents computer should recognize the behavior of a human player. The purpose of pattern recognition is to abstract relevant information from the game world and to construct concepts and to deduce patterns from this information. 204453: Pattern Recognition 5

Relations between the world, pattern recognition, and decision-making 204453: Pattern Recognition 6

Pattern Recognition in Computer Games Fighting reacts to enemy's frequent moves RTS: Real-Time Strategy remedies threats and strategizes 204453: Pattern Recognition Sports reads the match 7

Decision Making Levels Stance Towards Players Game Graphs PERSPECTIVE VIEWS OF PATTERN RECOGNITION IN COMPUTER GAMES 204453: Pattern Recognition 8

Decision Making Levels Strategical Tactical 204453: Pattern Recognition Operational 9

Strategical Long Period of Time Large Amount of Data – Inhabitants – Items – Events High Cost of a Wrong Decision Speculative: What-If Scenarios Offline: Background 204453: Pattern Recognition 10

Operational Concrete Atomatory Entities Reactive Short-Term Real-Time Online: in-the-Field Irrevocable Problems 204453: Pattern Recognition 11

Tactical connects between strategical and operational considers a group of entities and their cooperation made more frequently than strategical, pattern recognition has less time to use The quality cannot be as high as on upper level. 204453: Pattern Recognition 12

Stance Towards Players Enemy Ally Neutral 204453: Pattern Recognition 13

Enemy requires modus operandi of the player provides challenge 204453: Pattern Recognition 14

Ally accounts human perspective but not decision making system Synthetic Reconnaissance Officer – It reports on enemy movement. – It suggests effective counteractions. 204453: Pattern Recognition 15

Neutral Autonomous Camera Director – controls camera movement in sports games – dictates by television practice Referee – allows the play continue – interprets causality between offence and subsequent events Interface should adapt dynamically to the needs of a player. 204453: Pattern Recognition 16

Game Graph A story progresses linearly. A game provides an illusion of free will. Nodes – Game States Direct Arches – Actions The game properties can be analyzed through graph concepts (e.g., repetitiveness corresponds to cycles in the graph). 204453: Pattern Recognition 17

Outdegree The number of direct arches leaving a node The greater the outdegree means more freedom the player has. 204453: Pattern Recognition 18

Indegree The number of direct arches entering the node Uniqueness of a response can be measured as the indegree. 204453: Pattern Recognition 19

A linear (a story) allows no diversion 204453: Pattern Recognition 20

node Si has an outdegree of 2 204453: Pattern Recognition 21

node Sn has an indegree of 3. 204453: Pattern Recognition 22

Each action has now a unique response 204453: Pattern Recognition 23

References T. Kaukoranta, J. Smed, and H. Hakonen, Role of Pattern Recognition in Computer Games, Proceedings of the 2nd International Conference on Application and Development of Computer Games, pp. 189--94, Hong Kong SAR, China, 2003. https://www.nature.com/articles/ d41586-019-03298-6 204453: Pattern Recognition 24

Related Articles

Back to top button