Proportional Conflict Redistribution Rules for Information Fusion

TL;DR

This paper introduces five progressively complex Proportional Conflict Redistribution (PCR) rules for information fusion, improving conflict resolution accuracy and addressing limitations of previous methods like the WAO and Dempster's rule.

Contribution

The paper proposes new PCR rules that enhance conflict redistribution in information fusion, with specific improvements over existing rules such as PCR4, and discusses their theoretical properties.

Findings

PCR rules effectively redistribute conflicting masses

PCR4 improves upon minC and Dempster's rules

PCR1 aligns with WAO but lacks neutrality property

Abstract

In this paper we propose five versions of a Proportional Conflict Redistribution rule (PCR) for information fusion together with several examples. From PCR1 to PCR2, PCR3, PCR4, PCR5 one increases the complexity of the rules and also the exactitude of the redistribution of conflicting masses. PCR1 restricted from the hyper-power set to the power set and without degenerate cases gives the same result as the Weighted Average Operator (WAO) proposed recently by J{\o}sang, Daniel and Vannoorenberghe but does not satisfy the neutrality property of vacuous belief assignment. That's why improved PCR rules are proposed in this paper. PCR4 is an improvement of minC and Dempster's rules. The PCR rules redistribute the conflicting mass, after the conjunctive rule has been applied, proportionally with some functions depending on the masses assigned to their corresponding columns in the mass matrix. There are infinitely many ways these functions (weighting factors) can be chosen depending on the complexity one wants to deal with in specific applications and fusion systems. Any fusion combination rule is at some degree ad-hoc.