Un framework organisationnel pour la conception et l'implantation multi-agent de métaheuristiques (présentation courte)

Research in metaheuristics has recently evolved to new issues such as : simplicity, robustness and modularity of metaheuristics. Distributed Artificial Intelligence and particularly multiagent systems seem to be a promising field of research to tackle these new issues. In this paper we propose AMF, an Agent Metaheuristic Framework that aims at supporting the design and implementation of metaheuristics using an organizational and agent-oriented approach. This framework gives a model and a set of methodological guidelines to support the design of metaheuristics and facilitate the distribution and integration of artificial learning components. This paper also introduces an original metaheuristic called Coalition-Based Metaheuristic (CBM) that illustrates the use of AMF and artificial learning in metaheuristics.