一种基于置信最大熵模型的证据推理方法

D-S证据组合规则在处理高冲突信息时会得出与直觉相反的结论,这一直是D-S理论研究的热点.与相关理论优势互补是克服证据理论固有缺陷的有效方法之一.基于对最大熵原理和证据理论的研究,定义了辨识框架上的基本最大熵置信分配函数,并与经典的D-S组合规则及其改进方法相结合,给出了相关推理公式及基于置信最大熵模型.理论分析和实验表明,最大熵新证据的加入使非单焦元的基本置信赋值按比例重新分配给了单焦元,很好地处理了高冲突信息.

A combination rule of evidence theory based on brief max-entropy model

The combination rule of high conflict evidence is one of the important issues in D-S theory since it gives error results that contrary with the instinct.Combining relative theories into D-S theory is an effective way in dealing with the inherent defects of D-S theory.Based on the max-entropy and D-S theory,a basic belief max-entropy assignment function is proposed.Then a model based on max-entropy is obtained by combing Max-entropy with classic D-S theory and the alternative D-S theories,relevant rational formulas are presented.Theoretical analysis and experiments show that the incorporation of the new max-entropy evidence into D-S theory can better process high-conflict information by re-allocating the basic belief value of the non-single focal elements proportionally into the single focal elements.