A minimal model of communication for multi-agent systems

Classifier systems are rule-based systems dedicated to the learning of more or less complex tasks. They evolve thanks to a genetic algorithm toward a solution without any external help. When the problem is very intricate it is useful to have different systems, each of them being in charge with an easier part of the problem. The set of all the entities responsible for the resolution of each sub-task, forms a multi-agent system. Agents have to learn how to exchange information in order to solve the main problem. We define the minimal requirements needed by multi-agent classifier systems to evolve communication. We thus design a minimal model involving two classifier systems which goal is to communicate with each other. A measure of entropy that evaluates the emergence of a common referent between agents has been finalised. The minimal model applied to two sorts of classifier systems has shown promising results and let us think that this work is only the beginning of our ongoing research activity.