快速Fuzzy聚类方法的实现.pdfVIP

The 2nd Asia-Pacific Conference on Information Network and Digital Content Security Realization of Quick Fuzzy Cluster Method Shiqiang ZHANG1,2,Chunli WANG1,2,Ting WANG 1,2 1. Dept. of mathematics, Chongqing Medical University, Chongqing,China 2. Lab. of forensic medicine and biomedicine information, Chongqing Medical University, Chongqing,China Abstract: Cluster analysis in Statistics is a multi-dimensional analysis method. Traditional cluster analysis re- quires that the boundary between given objects is clear. Paper[1] analyzes different cluster processes in syste- matic cluster method. It turns out that there only one cluster process(maximum statistics method cluster process)will not interfere the primary information, no matter to target cluster(R cluster) or sample cluster(Q cluster); Other cluster processes will all increase Disturbance information or lose primary information. The key of cluster analysis is to build similar square. Once the boundaries between the given objects are unclear, the traditional cluster analysis is useless. The fuzzy cluster method introduced by paper [2] can overcome the flaw of system cluster. Based on paper [2], this paper introduces the calculation method of the degree of member- ship of fuzzy complementary set. It also enhances the fuzzy similar square and gives a feasible quick fuzzy cluster method. Keywords: Statistics;cluster analysis; system cluster; fuzzy cluster 快速Fuzzy 聚类方法的实现 1,2 1,2 1,2 张世强 ,王春丽 ,王婷 1.重庆医科大学数学教研室，重庆，中国，400016 2.重庆医科大学法医学与生物医学信息研究室，重庆，中国，400016 【摘要】统计学中的聚类分析(cluster analysis)是一种多元统计分析方法，传统聚类分析要求给定对象之 间的界限是清晰的。文[1]分析了系统聚类法中的各种聚类过程,发现无论是对指标聚类(R 型