ADAPTIVE BEHAVIOUR OF SOFTWARE AGENTS IN MULTI-AGENT COMPUTER GAME

• Main Authors: I. Yu. Sotnikov, I. V. Grigorieva
• Format: Article
• Language: English
• Published: Kemerovo State University 2014-11-01
• Series: Вестник Кемеровского государственного университета
• Subjects: artificial intelligence; computer games; agent-oriented approach; search path; machine learning
• Online Access: https://vestnik.kemsu.ru/jour/article/view/1063

The paper consider multi-agent games, their structure, software presentation of agents. Games present different genres, have different rules and structure of the world. As examples of use of games as a platform for the study of artificial intelligence, agents are considered with the use of search algorithms path and learner agents. Algorithm A* and Lee algorithm were taken as path search algorithms. To implement learner agents machine-learning techniques were selected such as reinforcement learning and artificial neural network taught by a genetic algorithm.