A size-insensitive integrity-based fuzzy c-means method for data clustering
作者：
Highlights：
• We propose a new conditional FCM method based on “integrity” and size-ratio of clusters.
• The proposed method can significantly alleviate the “cluster-size sensitivity” problem.
• The proposed method has much bigger tolerance for the “distance” between clusters.
• The proposed method has more flexibility of selecting the initial cluster centers to keep the clustering method work successfully.
• Our method has much higher clustering accuracy than FCM and csiFCM for clustering datasets containing unbalanced clusters.
摘要
•We propose a new conditional FCM method based on “integrity” and size-ratio of clusters.•The proposed method can significantly alleviate the “cluster-size sensitivity” problem.•The proposed method has much bigger tolerance for the “distance” between clusters.•The proposed method has more flexibility of selecting the initial cluster centers to keep the clustering method work successfully.•Our method has much higher clustering accuracy than FCM and csiFCM for clustering datasets containing unbalanced clusters.
论文关键词：Fuzzy c-means,Cluster size insensitive,Integrity,Compactness,Purity,Data clustering
论文评审过程：Received 4 September 2012, Revised 19 October 2013, Accepted 28 November 2013, Available online 12 December 2013.
论文官网地址：https://doi.org/10.1016/j.patcog.2013.11.031
