DBE: Dynamic belief entropy for evidence theory with its application in data fusion

被引：6

作者：

Deng, Jixiang ^{[1
]}

Deng, Yong ^{[1
,2
,3
,4
]}

机构：

[1] Univ Elect Sci & Technol China, Inst Fundamental & Frontier Sci, Chengdu 610054, Peoples R China

[2] Shaanxi Normal Univ, Sch Educ, Xian 710062, Peoples R China

[3] Japan Adv Inst Sci & Technol, Sch Knowledge Sci, Nomi, Ishikawa 9231211, Japan

[4] Swiss Fed Inst Technol, Dept Management Technol & Econ, CH-8093 Zurich, Switzerland

来源：

ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE | 2023年 / 123卷

基金：

中国国家自然科学基金;

关键词：

Dempster-Shafer evidence theory; Uncertainty measurement; Belief entropy; Dynamic belief entropy; Data fusion; TOTAL UNCERTAINTY MEASURE; INFORMATION; DISTANCE;

D O I：

10.1016/j.engappai.2023.106339

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

Belief entropy is an effective uncertainty measurement in Dempster-Shafer evidence theory. However, the weight ratio between discord and non-specificity in the belief entropy is static and cannot be further modified according to different environments. To overcome this issue, this paper proposes dynamic belief entropy (DBE), which is a generalization of belief entropy by introducing a dynamic parameter. Compared with belief entropy, DBE can be flexibly modified based on the dynamic parameter, so as to improve the performance of measuring uncertainty in different environments. Besides, some properties of DBE are presented and illustrated with examples. Also, we design a dynamic data fusion method based on DBE. Compared with the existing methods, the proposed method utilizes DBE-based dynamic techniques, thereby enhancing the classification performance. Moreover, to illustrate the general applicability, the proposed method is verified on classification problems. The experimental results show that the proposed method outperforms the existing methods with a classification accuracy of 95.93% and an F1 score of 96.08%, demonstrating the effectiveness of our method.

引用

页数：13