A new base function in basic probability assignment for conflict management

He, Yuanpeng; Xiao, Fuyuan

doi:10.1007/s10489-021-02525-w

Cited by 5 publications

(2 citation statements)

References 67 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…• Xiao et al's base belief function [60] This base belief function approach is introduce to decrease conflicts among evidence sets and thus to diminish the loss appear in the process of combination. This base belief function consider the cardinality of the proposition to suitably changes the BPA distributed to specific proposition.…”

Section: Existing Base Belief Functionmentioning

confidence: 99%

See 1 more Smart Citation

Leveraging a New Base Belief Function for Conflict Data Management and Decision Support

Dutta,

Shome

2023

Preprint

View full text Add to dashboard Cite

The Dempster-Shafer theory (D-S theory) finds wide-ranging applications, including pattern recognition and decision making. When paradoxes exists in multi-source information system due to unfavorable environment, unification cannot be reliably and effectively performed by traditional D-S theory. In light of this, a fresh perspective is taken to construct a base belief function for conflict data fusion within the D-S theory. In this novel approach, a new base belief function is introduced to modify the original basic probability assignment (BPA) prior to applying the classical Dempster combination rule (DCR) in a closed world scenario. Instead of assigning initial belief to the entire power set of the Frame of Discernment(FOD), the proposed method assigns base belief to individual events. This targeted approach aims to reduce the existing uncertainty within the FOD. To generate more intuitive outcomes, the proposed method combines the base belief function with the original BPA, thus incorporating all the pertinent information contained in the evidence set (BOE). Compared to other alternative methods, the proposed approach demonstrates consistent and logical outcomes that align with the real world. Furthermore, it exhibits a high rate of convergence and lower computational complexity. An advantageous aspect of the proposed method is its utilization of maximum information from the evidence set, not limited to an average operation between the base belief function and the initial BPA. This comprehensive integration not only resolves conflicts within the evidence set but also generates effective outcomes. To validate the efficaciousness of the proposed method, several numerical examples are presented, along with its application to classification problems. These instances substantiate the effectiveness of the approach in practice.

show abstract

Section: Existing Base Belief Functionmentioning

confidence: 99%

“…It is worth noting that the DCR proves ineffective in handling all four paradoxes. Consequently, we have chosen to evaluate the performance of seven established methods, which were respectively proposed by Yager[38], Sun et al[61], Wang et al[58], Jing et al[59], Xiao et al[60], Murphy…”

mentioning

confidence: 99%