SFERES : Un framework pour la conception de systèmes multi-agents adaptatifs

We present a framework for artificial evolution and multi-agent simulation for the design of adaptive multi-agent systems by means of evolutionnary algorithms. In the first part, the concepts and techniques of existing evolutionnary algorithms are introduced. The second part describes the framework and its main components, the general structure of SFERES, the design directions and some extensions. We conclude with applications taking advantage of the environment and facilities of SFERES together with some development perspectives. MOTS-CLÉS : framework, apprentissage multi-agent, évolution artificielle, simulation multi-agent