Negation of Basic Belief Assignment in Multisource Information Fusion on Dempster-Shafer Theory With Applications in Pattern Classification.

Abstract

In the construction of complex decision-making systems, which often involve uncertainties from multiple sources of information, effectively expressing the uncertainty of information remains an unresolved issue. Therefore, on the basis of Dempster-Shafer theory (DST), we define the negation of basic belief assignment (NBBA) via the difference measure of focal elements, investigate this problem from a negation perspective, and apply the NBBA in decision-making. First, we propose a kernel entropy adaptive focal element generation method (KEAFG) for constructing expressive basic belief assignments (BBAs). On this basis, we effectively construct NBBAs. Furthermore, to provide a theoretical foundation for decision-making based on the NBBA, we analyze and discuss the theoretical properties of the NBBA. We subsequently propose a multisource information fusion classification method based on the NBBA (MSIF-NBBA). This method involves joint decision-making on the BBAs obtained by the KEAFG and the NBBAs obtained by the NBBA to obtain classification results. MSIF-NBBA has been validated on several datasets, including CIFAR-10, MNIST, Fashion-MNIST, Iris, and Heart. Experimental results on multiple datasets demonstrate that the proposed method significantly outperforms state-of-the-art methods. The proposed method effectively reduces uncertainty in multisource information fusion and improves decision-making. The code is available at https://github.com/LeonidMs-L/MSIF-NBBA.