基于紧密度的模糊加权kNN数据分类方法

Data Classification Method of Fuzzy Weighted k-Nearest Neighbor Based on Affinity

模糊k-最近邻(fuzzy k-nearest neighbor,FkNN)及其改进的分类方法忽略了样本存在分布不均匀以及噪声样本的情况,不能充分体现每个类样本特征的差异性,影响了分类的准确率.为此,提出了一种基于紧密度的模糊加权kNN数据分类方法.首先基于样本间紧密度计算样本的隶属度;然后根据特征的模糊熵值分别计算每个类样本特征的权重,并使用加权欧氏距离确定近邻训练样本;最后根据待分类样本所属的每个类别的隶属度确定其类别.对UCI多个数据集的实验结果表明该方法是有效的.

In sample classification, the fuzzy k-nearest neighbor （FkNN） method and the associate improved classification algorithms ignore the uneven distribution of samples and the noise samples, thus are unable to reflect the differences of class sample features, resulting in the low classification accuracy. In order to overcome the limitations, a fuzzy weighted k-nearest neighbor data classification method based on affinity is proposed in this paper. Firstly, the membership of samples is calculated based on affinity among samples. Then, the feature weights of class samples are determined by the fuzzy entropy values, and k-neighbors are selected according to the weighted Euclidean distance. Finally, the samples will be classified according to the fuzzy membership of the samples belong to each class. The experimental results on the UCI datasets show that the proposed method is effective.