一种基于D-S证据理论的状态识别方法
A Method of State Recognition Based on Theory of Evidence
-
摘要: 通过距离函数求出待识别样本属于各类别的似然函数.对难识别和易识别错误的样本,通过新证据的引入,合成2批或多批证据下的似然函数,确定样本的类别,从而减少了非典型样本的误识率,提高了识别的正确率.Abstract: By using the method of state recognition,a plausibility function which describes the resemblance between the sample and each class can be obtained through distance function.For the marginal samples,a new plausibility function based on two or several groups of evidences can be determined according to plausibility function.This method provides a better framework for modeling the human being's reasoning process. The result proves that the rate of recognition correctness is satisfactory.