证据理论k-NN规则中确定相似度参数的新方法

A New Approach to Determine the Similarity Parameters in Evidence-Theoretic k-NN Rule

本文提出了一种确定证据理论k NN分类规则中相似度参数的新方法.对于一个模式识别问题,我们首先为每一模式类求得一个参考最近邻距离,使其在最小错误率意义下将训练样本集中属于该模式类的样本与其他样本分离,然后根据所得参考最近邻距离计算相似度函数参数.该方法在训练集比较小、样本非高斯分布条件下仍然能够计算出比较准确的参数,使得相应的分类错误率较小,而且时间复杂度比L .M .Zouhal的方法低约4 8倍.

This paper presents a new approach to determine the similarity parameters in the Evidence-Theoretic k-NN Classification Rule. Given a pattern recognition problem,we first compute a reference nearest neighbor distance to separate samples of one class from other samples with least error rate,and then calculate the similarity parameters based on the obtained distance. Under the condition of small scale samples with non-gaussian distribution,the proposed method can get more suitable parameters and thus reduce classification error rate.Furthermore,its computation complexity is 4-8 times lower than that of L.M.Zouhal's method.