Title :
A modified evidence combination approach based on ambiguity measure
Author :
Han, Deqiang ; Han, Chongzhao ; Yang, Yi
Author_Institution :
Inst. of Integrated Autom., Xian Jiaotong Univ., Xian
fDate :
June 30 2008-July 3 2008
Abstract :
Dempster rule of combination is an effective method in evidence reasoning; however, the original combination rule may produce counterintuitive results when there are severe conflicts among evidences due to the discarding of contradictory mass assignments. In this paper a modified combination rule is proposed based on ambiguity measure (AM), which can describe the evidencepsilas degree of uncertainty. Weight factors are generated based on the AM of evidences to reallocate conflicting mass assignments. The combination results based on the proposed approach can be more reasonable. Experiments and examples provided show the efficacy and validity of the proposed approach. Comparisons with other modified evidence combination rules are also provided.
Keywords :
case-based reasoning; knowledge based systems; probability; Dempster combination rule; ambiguity measure; evidence reasoning; mass assignments; modified evidence combination approach; Evidence theory; ambiguity measure; combination rule; conflicts redistribution; information fusion;
Conference_Titel :
Information Fusion, 2008 11th International Conference on
Conference_Location :
Cologne
Print_ISBN :
978-3-8007-3092-6
Electronic_ISBN :
978-3-00-024883-2