一种K-均值脸谱图聚类新算法

Novel clustering algorithm using K-means Chernoff faces

Chernoff脸谱图简单,类似卡通画,能图形化地表示多元数据。但脸谱图聚类算法具有主观性的巨大的对比工作量,脸谱特征分配困难。因此,本文提出一种新的脸谱图聚类算法,它合并了K均值聚类或模糊G均值聚类算法。IRIS和蔬菜油数据集的实验结果表明新算法优于传统的聚类算法。

Chernoff faces are simplified, cartoon-like faces that can be used to graphically display complex multivariate data. Chernoff faces are clustering algorithms, which group together similar faces. Some disadvantage of Chernoff faces for clustering are the subjectivity, huge comparison workload and difficulty in assignment of facial features. So, a novel Chernoff face clustering algorithm is proposed, which combines with K-means or FCM clustering algorithms. The experiment results of IRIS data and vegetable oil data indicate that the proposed new algorithm is superior to traditional clustering algorithms.