Papers
Topics
Authors
Recent
Search
2000 character limit reached

A correlation coefficient of belief functions

Published 16 Dec 2016 in cs.AI | (1612.05497v2)

Abstract: How to manage conflict is still an open issue in Dempster-Shafer evidence theory. The correlation coefficient can be used to measure the similarity of evidence in Dempster-Shafer evidence theory. However, existing correlation coefficients of belief functions have some shortcomings. In this paper, a new correlation coefficient is proposed with many desirable properties. One of its applications is to measure the conflict degree among belief functions. Some numerical examples and comparisons demonstrate the effectiveness of the correlation coefficient.

Citations (230)

Summary

  • The paper presents a novel correlation coefficient that effectively measures and manages conflict in Dempster-Shafer belief functions.
  • It rigorously demonstrates the coefficient's superiority and sensitivity over traditional methods through detailed numerical comparisons.
  • The methodology offers practical applications in sensor fusion, decision support systems, and fault diagnosis under uncertainty.

A Correlation Coefficient of Belief Functions: A Technical Examination

The paper "A Correlation Coefficient of Belief Functions" by Wen Jiang explores a critical aspect of Dempster-Shafer (D-S) evidence theory, specifically addressing the challenge of conflict management. D-S theory is utilized widely for decision-making under uncertainty, yet it struggles in scenarios where conflicting evidence arises. This paper introduces a novel correlation coefficient for belief functions, intended to measure and manage these conflicts more efficaciously than existing methodologies.

Core Contributions

The primary contribution of this work is the formulation of a new correlation coefficient that aims to overcome the limitations of current coefficients used in D-S theory. This new coefficient considers both non-intersection and differences among focal elements of belief functions, providing a more comprehensive measure of conflict. Its utility is demonstrated through numerical examples that highlight its effectiveness relative to notable existing methods such as Jousselme’s evidence distance and Song et al.'s correlation coefficient.

Theoretical Foundation

The paper establishes a theoretical framework that outlines desirable properties for correlation coefficients in the context of belief functions. These properties include symmetry, bounds within [0,1], and specific conditions for total conflict and total agreement. The proposed correlation coefficient satisfies these properties, distinguishing itself with mathematical proofs that ensure its robust performance in conflict measurement.

Numerical Demonstration

Through a series of illustrative examples, the paper compares the performance of the proposed correlation coefficient against existing methods. Key numerical results indicate that the new coefficient reliably identifies both total agreement and total contradiction between evidence sets. For instance, in scenarios where two belief functions are identical, the proposed coefficient correctly computes a conflict measure of zero, unlike the classical conflict coefficient and some other measures.

Implications and Future Directions

The introduction of this correlation coefficient could enhance practical applications of D-S theory, particularly in fields requiring reliable uncertainty quantification such as sensor fusion, decision support systems, and fault diagnosis. By providing a more sensitive and accurate conflict measure, the approach could lead to improvements in systems traditionally challenged by conflicting data.

Theoretical implications extend to evidence theory itself, suggesting new avenues for refining combination rules and handling partial beliefs more effectively. Future research may explore extensions to the proposed coefficient, including its integration with machine learning techniques for dynamic conflict management in real-time systems.

Conclusion

This paper presents a significant advancement in addressing a long-standing challenge in Dempster-Shafer evidence theory—the management of conflicting evidence. By introducing a new correlation coefficient grounded in sound theoretical principles and validated through thorough numerical analysis, the research offers a promising tool for both theoretical exploration and practical application in evidence-based decision-making frameworks. While further exploration and validation in diverse scenarios remain necessary, the proposed method holds potential for expanding the applicability and reliability of D-S theory in complex environments.

Paper to Video (Beta)

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Authors (1)

Collections

Sign up for free to add this paper to one or more collections.