Apply Cultural Algorithm to Co-Evolution of Multi-Agent System

• Main Authors: Hsing-Chang Lai, 賴信彰
• Other Authors: Chien-Chang Hsu
• Format: Others
• Language: en_US
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/19393779292528502803

碩士 === 輔仁大學 === 資訊工程學系 === 94 === In this paper, we proposed a framework called Cultural Coevolutionary Multi-Agent Model (CCMAM) which applies cultural algorithm to the coevolutionary of 3APL-based Multi-Agents platform. Our purpose is to develop a coevolvable agent group model which includes two features; to improve the capability of agent or agent group, to improve the cooperation between agents, via coevolvable cultural algorithm model. The application domain is a pursuit-evasion game includes three groups of teams: pursuers, evaders and obstacles. Result indicate that the cultural algorithm applied within both competitive and cooperative coevolution process at the same time yields ascendant performance comparative to apply only either competitive or cooperative approach.