The Applications of Dempster-Shafer Theory in Multiple Classifiers Combination
Multiple Classifiers Combination is to utilize distinguishedclassifiers to resolve the same classification problem as a single classifier does, which can improve performance and generalization capability.This paper presents an overview of multiple classifiers combination based on Dempster-Shafer Theory. Mass functions generation is the most difficult step in practical. In this paper the approaches to generating mass functions in multiple classifiers combination are discussed which include the approaches based on global classification performance, based on class-wise classification performance and based on outputs measurement information of member classifiers. Mass functions generation is application-dependent. There is still no unified approach. And no approach is considered to be the best for all applications. The available approaches have not only many strengths and but also some weakness. With the development of the researches on evidence theory, especially the emergency of more reasonable generation approaches of mass functions, the performance of classifiers combination can be expected to be better. The utilization range of evidence theory can be extended in applications of multiple classifiers combination.
Evidence Theory Multiple Classifiers Combination Mass function
Deqiang Han Chongzhao Han
Institute of Integrated Automation Xian Jiaotong University Xian, Shaanxi, China PR
国际会议
The International Colloquium on Onformation Fusion 2007(2007年国际信息融合研讨会)
西安
英文
398-403
2007-08-22（万方平台首次上网日期，不代表论文的发表时间）