A complex belief [formula omitted] divergence in complex evidence theory and its application for pattern classification

Abstract

Complex evidence theory (CET) is crucial in modeling uncertain information in the complex domain. With the development of the research on CET, how to measure the conflict between complex basic belief assignments (CBBAs) accurately remains an open issue. Thus, we propose the complex belief χ2 divergence (called CBχ2 divergence) tending to solve conflict management issues. The proposed complex belief χ2 divergence which considers the quantum interference effects is a generalized model of traditional belief χ2 divergence. Furthermore, the proposed complex belief χ2 divergence satisfies the properties of symmetry, non-degeneracy and continuity. Besides, an algorithm for decision-making based on the complex belief χ2 divergence is proposed. This method is applied in several datasets from UCI machine learning repository including the iris dataset, the seeds dataset and the heart disease dataset for pattern recognition. The result shows the proposed method has better classification accuracy compared with the related methods.