A Self-adaptive Feature Weighted FCM Clustering Algorithm Based On Entropy

摘要：

In view of fuzzy c-means algorithm have limitation when applied to clustering analysis, this paper propose a new algorithm-self-adaptive feature weighted FCM algorithm. Through pre-processing mechanism, new algorithm can self-adaptive obtain cluster number and initial centers according to data set distribution characteristic. To consider particular contribution of different feature in clustering, multi-factors weight portion method based on message entropy is employed to determine weight of each feature, and feature weight is introduced in clustering criteria of FCM algorithm. Applied the algorithm to real data set, experimental results illustrate that the algorithm can self-adaptive determine cluster number, and clustering results have higher stability and higher accuracy.

关键词： FCM clustering pre-processing algorithm inter-cluster entropy weighted fuzzy clustering

作者: Ming-Xing Ye Rong Wei Zheng-Wu Yuan

作者单位: Sino-Korea Chongqing GIS Research Center, College of Computer Science and Technology,Chongqing University of Posts and Telecommunications, China

会议类型: 国际会议

会议名称: The 8th Asian Symposium on Geographic Information Systems from a Computer Science & Engineering Viewpoint(ASGIS 2010)(第八届亚洲地理信息系统国际学术研讨会)

会议地点: 重庆

会议语种:英文

页码: 172-175

在线出版日期: 2010-04-22（万方平台首次上网日期，不代表论文的发表时间）