Study of Multi-agent Self-Organization System

Self-organization of the intelligent agents is accomplished because each agent models other agents by observing their behavior. Agents have belief, not only about environment, but also about other agents. To study the proposed intelligent agent´s learning and self-organizing abilities, in this paper, we explain the structure of an agent, which is designed by a Bayesian network and an influence diagram, and then examine a multi-agent organization system and the bi-directional learning feature of the proposed multi-agent self-organizing system. Finally, we present the system representation of the decision-theoretic intelligent agent design.