Generalized Belief Entropy and Its Application in Identifying Conflict Evidence

Abstract

Dempster-Shafer evidence theory has wide applications in many fields. Recently, A new entropy called Deng entropy was proposed in evidence theory. Some scholars have pointed out that Deng Entropy does not satisfy the additivity in uncertain measurements. However, irreducibility may have a huge effect. The derived entropy from complex systems is often irreducible. Inspired by this, generalized belief entropy is proposed. The belief entropy implies the relationship between Deng entropy, Rényi entropy, Tsallis entropy. In addition, numerical examples demonstrate the flexibility of the proposed Rényi-Deng (R-D) entropy to measure the uncertainty of basic probability assignment (BPA). Finally, a method for identifying contradictory evidence based on Rényi-Deng (R-D) entropy is proposed. The experiment show the effectiveness of the proposed method.