Title :
A novel algorithm for coalition formation in Multi-Agent Systems using cooperative game theory
Author :
Ghazikhani, Adel ; Mashadi, Habib Rajabi ; Monsefi, Reza
Author_Institution :
Eng. Dept., Imam Reza Univ., Mashhad, Iran
Abstract :
We propose a novel algorithm for coalition formation in Multi-Agent Systems (MAS). Coalition Formation (CF) is a type of cooperation in MAS. This cooperation is expressed with coalitions of agents. In literature, CF has been viewed as an optimization problem. Game theory is a subject that is meant to solve optimization problems where multiple players are affecting the decision vector. In this work, cooperative game theory has been used to solve for an optimized CF. An algorithm based on Shapley value, which is a concept concerning cooperative game theory has been utilized in this research work, due to its capabilities of assigning contribution to players. We evaluated our algorithm using different settings of multiple agents and tasks. The proposed algorithm has shown satisfactory results.
Keywords :
Artificial intelligence; Bayesian methods; Convergence; Evolutionary computation; Game theory; Learning; Multiagent systems; Problem-solving; Scalability; Uncertainty; Shapley value; coalition formation; cooperative game theory; multi-agent systems;
Conference_Titel :
Electrical Engineering (ICEE), 2010 18th Iranian Conference on
Conference_Location :
Isfahan, Iran
Print_ISBN :
978-1-4244-6760-0
DOI :
10.1109/IRANIANCEE.2010.5507017