一种带有显著特征项的模糊聚类算法  

An Algorithm with Significant Feature for Fuzzy Clustering

作　　者：张东生[1] 张缨[1] 

机构地区：[1]河南大学计算中心,河南开封475004

出　　处：《河南大学学报（自然科学版）》2011年第2期184-187,共4页Journal of Henan University:Natural Science

基　　金：国家863高科技计划资助项目(2007AA01Z478);河南省教育厅自然科学研究指导计划资助项目(2008B520003)

摘　　要：模糊聚类分析可将特征相近的对象聚集为一类,但不直接显示聚类依据.改进算法根据对象特征项对聚类分析的贡献率,提出"显著特征项"的概念,进而得出聚类的主要依据.仿真结果表明,算法的聚类依据明确,数学基础可靠,计算过程规范.Fuzzy clustering analysis aims to integrate the object ingredients of similar features without demonstrating the evidence for clustering.This paper gives a notion of what is called ＂significant feature＂ in improving the method according to the contribution rate of the object features for clustering,and the main evidence of clustering can be figured out.Result of simulation proves that the evidence of clustering is valid and of practical use due to its reliability in computing method and the regularity in the process of computation.

关 键 词：模糊聚类 相似矩阵 传递闭包 显著特征项 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]