截集型特征加权模糊C-均值聚类算法

Clustering Algorithm of Sectional Set Feature-weighted Fuzzy C-means

已有的特征加权型模糊C-均值(WFCM)聚类算法可以有效地提取数据的相关特征,WFCM存在的主要问题是收敛速度慢和对噪声敏感。借助模糊集的截集方式对WFCM的隶属度值进行修改,提出截集型特征加权模糊C-均值聚类算法:SWFCM。SWFCM不仅具有良好的特征提取能力,而且具有收敛速度快和对噪声稳健的优点。实验结果表明,SWFCM的总体性能优于原有的WFCM聚类算法和截集模糊C-均值聚类算法。

Although the clustering algorithm of feature-weighted fuzzy C-means（WFCM）can effecively extract the related features of data, it still has some shortages such as slow convergence and sensitive to noise. The clustering algorithm of sectional set feature-weighted fuzzy C-means （SWFCM）is presented by revising the membership function of WFCM by sectional set. In comparison with WFCM, SWFCM not only can extract the features of data set, but also has the advantages of fast convergence and robust to noise. The experimental results show that SWFCM has super performance over original WFCM clustering algorithm and sectional set fuzzy C-means clustering algorithm.