Babak Khosravifar

Trust and Reputation in Multi-agent Systems: Theoretical and Experimental Study of Secure Multi-agent System Design with Rational Intelligent Agents Babak Khosravifar

Artificial Intelligence has become one of the most fundamental areas of computer science research. One line of research in artificial intelligence is associated with coordination of intelligent agents. Coordination takes place in multi-agent environments where distributed agents have limited knowledge about their surrounding environment; therefore, they continuously ask other agents to obtain required information. In general, the growing popularity of agents requires systematic coordination management and reputation system that enable agents to decide about their interacting partner and overall acting attitude. Using the reputation system, agents choose reliable agents to interact with. Here, the question arises that how the reputation mechanism helps agents to make the most prudent decisions that yield the best outcome. More specifically, agents need to use a decision-theoretic reasoning algorithm to optimize their decisions under uncertainty. In multi-agent systems, the reputation value is highly competitive because it is used as the beliefs of agents about one another. Thus, a carefully designed mechanism is required to maintain accuracy of this parameter.