A Novel Possibilistic Fuzzy C-Means Clustering
Zhou Jian-jiang · Dianzi xuebao · 2008
Fuzzy c-means(FCM) algorithm is sensitive to noises and possibilistic c-means(PCM) algorithm is very sensitive to the initialization of cluster centers and generates coincident clusters.Based on combination of FCM and PCM,possibilistic fuzzy c-means clustering(PFCM) overcomes these shortcomings.However,PFCM must run FCM to compute the parameters beforehand.A novel PCM is proposed and it computes the parameters with covariance matrix to judge the compactness of data sets.Furthermore,the proposed PCM need not run FCM beforehand.A novel PFCM is obtained based on the novel PCM and FCM.The novel PFCM need not run FCM beforehand to compute the parameters,so it reduces the clustering time.The experimental results with data sets show that our proposed algorithm can produce fuzzy membership values and typicality values simultaneously,less clustering time and better clustering accuracy.