A novel complex evidence distance and multisource information fusion for pattern classification
Qiwen Pang, Yangyang Zhao, Fuyuan Xiao · Journal of Control and Decision · 2025
Complex evidence theory serves as an efficient method for fusing complex basic belief assignments, yielding reasonable outcomes. In complex evidence theory, conflict management remains an ongoing topic. As quantum theory develops, it increasingly acts on the problem of uncertainty. Herein, we introduce a new complex evidence distance. This metric takes the quantum interference phenomenon among focal elements into consideration, providing a more accurate degree of conflict intensity among complex basic belief assignments. This new complex distance formula maintains the basic distance property, including non-negativity, non-degeneracy, symmetry, and trigonometric inequalities. Moreover, we propose a novel multisource information fusion algorithm for pattern classification on the basis of the newly defined complex evidential distance. Finally, we apply the proposed algorithm to Iris classification, medical diagnosis and occupancy detection in real-world scenarios to validate its effectiveness. The outcomes of these experiments affirm the potency of the proposed complex evidence distance in addressing classification tasks with genuine complexities.