[1]马志民,陈汉武,张 军.一种适用于不规则分布数据的混合聚类算法[J].南京师范大学学报(工程技术版),2006,06(01):057-60.
　MA Zhimin,CHEN Hanwu,ZHANG Jun.A Hybrid Clustering Algorithm for Irregular Distributed Data[J].Journal of Nanjing Normal University(Engineering and Technology),2006,06(01):057-60.
点击复制

一种适用于不规则分布数据的混合聚类算法
分享到：

南京师范大学学报（工程技术版）[ISSN:1006-6977/CN:61-1281/TN]

卷:
06卷
期数:
2006年01期
页码:
057-60
栏目:
出版日期:
2006-03-30

文章信息/Info

Title:
A Hybrid Clustering Algorithm for Irregular Distributed Data
作者:
马志民1;  2;  陈汉武1;  张  军1
1. 东南大学计算机科学与工程系, 江苏南京210096
2. 江西省信息中心, 江西南昌330046
Author(s):
MA Zhimin; CHEN Hanwu; ZHANG Jun
1.Department of Computer Science and Engineering,Southeast University,Nanjing 210096,China;2.Jiangxi Province Information Center,Nanchang 330046,China
关键词:
模糊划分;  FCM;  层次聚类;  模糊度量
Keywords:
fuzzy partition;  FCM;  h ierarchical c lustering;  fuzzy m easure
分类号:
TP301.6
摘要:
作为数据挖掘的一项重要技术,聚类分析具有广泛的应用领域.同时,聚类也是数据挖掘领域中一个相对比较困难的问题.在聚类算法中,基于模糊划分的FCM算法是一种重要的算法.和其它的算法相比,FCM算法具有计算简单、运算速度快,且有比较直观的几何意义的优点,因此在图像处理、模式识别等领域得到了广泛的应用.和所有的c均值算法一样,FCM算法也是只用类中心来表示类,这样只是适合球状类型的簇.本文在目前FCM算法研究的基础上,讨论了传统FCM算法在原型初始化上的局限性.提出一种基于层次凝聚的改进算法,使之能够适用于不规则分布的数据.
Abstract:
C luster ing analysis, as an important techno logy of datam in ing, has aw ide rang e o f application areas, but at the same time, cluster ing is a ra ther difficu lt problem in da tam in ing a rea. In comm on c lustering a lgo rithm s, FCM based on fuzzy division is one o f the im portant a lgo rithm s. Com pared w ith o ther a lgor ithm s, FCM has m any advan tag es such as simp le computation, rap id speed and an intuitive geom etr ic significance. So it has aw ide app lication in m any a reas such as im age processing and pattern recognition. A s many c-m eans a lgo rithm s, FCM denotes class only by c lass center, wh ich can only fit to sphere- like type o f cluster. This d issertation discusses the lim itations o f trad-i tional FCM algorithm in in itia liza tion o f pro totype. The paper presents a new a lgo rithm based on h ierarch ica l c lustering wh ich can be applied to the irregular distributed data.

参考文献/References:

[ 1] JIAW EIHAN, M ICH ELINE KAM BER. DataM in ing: Concept and Techniques [M ]. San Fransisco: M organ K aufm ann Publishers, Inc, 2001: 223-239.
[ 2] 高新波. 模糊聚类分析及其应用[M ]. 西安: 西安电子科技大学出版社, 2004: 92-97.
GAO X inbo. Fuzzy C lusterAna lysis and its Application[M ]. X i’an: X id ian Un iversity Pub lisher, 2004: 92-97.
( in Chinese)
[ 3] ESTE J, KR IEGEL H P, SANDER J, et a.l A dens ity-based algor ithm for d iscov ering c lusters in large spatia l databases w ith no ise[ J]. Proc KDD, 1996: 226-231.
[ 4] PAB ITRA M ITRA C A, MURTHY, SANKAR K PAL. Density-based mu ltiscale data condensation[ J]. IEEE Trans PAM I, 2002, 6( 24): 734-747.
[ 5] CHAUDHURI D, CHAUDHURI B B. A nove l m ultiseed nonh ie rarchical da ta cluster ing technique [ J]. IEEE T rans SMC, 1997, 10( 27): 871-877.

备注/Memo

备注/Memo:
基金项目: 国家自然科学基金资助项目( 90412014) .
作者简介: 马志民( 1975-) , 讲师, 主要从事数据挖掘等方面的研究, E-ma il:m zm seu@ sin a. com
更新日期/Last Update: 2013-04-29