Title :
Combination rule for belief functions based on improved measure of conflict
Author :
Peng, Ying ; Shen, Huairong
Author_Institution :
Co. of Postgrad. Manage., Acad. of Equip. Command & Technol., Beijing, China
Abstract :
Aiming at the invalidation of Dempster´s rule for combining belief functions with high conflict, this paper proposes a new combination rule. Firstly, the limitation of classical conflict measure is pointed out, and improved measure for degree of conflict is presented and proved. Based on this improved measure, a new combination rule which is composed of three steps is proposed. The first step is to find the most believable belief function, and the second step is to discount belief functions according to degree of conflict, finally, the belief functions is combined using Dempster´s rule. The proposed rule can combine belief functions with high conflict as well as belief functions with no conflict well. Two illustrative examples show the effectiveness and convergence of the proposed rule.
Keywords :
belief maintenance; inference mechanisms; uncertainty handling; Dempster rule; belief function; combination rule; conflict degree measure; Convergence; Data models; Electronic mail; Finite element methods; Lead; Reliability; Uncertainty; belief functions; combination rule; conflict;
Conference_Titel :
Information Theory and Information Security (ICITIS), 2010 IEEE International Conference on
Conference_Location :
Beijing
Print_ISBN :
978-1-4244-6942-0
DOI :
10.1109/ICITIS.2010.5689757