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

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.

Automated summary

This paper proposes dynamic belief entropy (DBE) which improves the measurement of uncertainty by introducing a dynamic parameter. A dynamic data fusion method based on DBE is designed, enhancing the classification performance. Experimental results demonstrate the effectiveness of the proposed method.