北京航空航天大学学报 ›› 2006, Vol. 32 ›› Issue (06): 747-750.

• 论文 • 上一篇    

基于核方法的聚类算法及其应用

纪秋颖1, 林健2   

  1. 1. 五邑大学 管理学院, 江门 529020;
    2. 北京航空航天大学 经济管理学院, 北京 100083
  • 收稿日期:2006-03-22 出版日期:2006-06-30 发布日期:2010-09-20
  • 作者简介:纪秋颖(1962-), 女, 吉林吉林人, 副教授, 现为北航在职博士生,jiqiuying@126.com.
  • 基金资助:

    国家自然科学基金资助项目(70471074);广东省科技攻关资助项目(2004B36001051)

Clustering algorithm based on kernel methods and its application

Ji Qiuying1, Lin Jian2   

  1. 1. School of Management, Wuyi University, Jiangmen 529020, China;
    2. School of Economics and Management, Beijing University of Aeronautics and Astronautics, Beijing 100083, China
  • Received:2006-03-22 Online:2006-06-30 Published:2010-09-20

摘要: 在分析核方法的核心概念基础上,提出了一种基于核方法的聚类算法.通常,传统聚类算法只在数据特征差异较大时才有效,当数据特征差异较小时,很难取得较好的聚类效果.引入核函数,将原始数据由数据空间映射到特征空间,在特征空间中进行聚类.核函数的非线性映射使得原始数据的特征更完整地显现出来,从而能够更客观准确地聚类.与传统聚类方法相比,该方法聚类结果更客观有效.以16组实际数据为例,将该方法应用于数据分类研究中,聚类结果表明了该方法的可行性和有效性,从而为数据分类提供了一种新的可行方法.

Abstract: Based on the analysis of the core concepts of the kernel methods, a clustering algorithm based on kernel methods was put forward. In general, traditional clustering algorithms are suitable to implement clustering only if the feature differences of data are large. If the feature differences are small and even cross in the original space, it is difficult for traditional algorithms to cluster correctly. By using kernel functions, the data in the original space was mapped into a high-dimensional feature space, in which more features of the data were exposed so that clustering could be performed efficiently. Compared with the traditional clustering methods, this clustering method had superiorities in dealing with the nonlinear data, which made its clustering result more objective and valid. This method was applied to the classification of 16 groups of data, and results show the feasibility and effectiveness of the kernel clustering algorithm.

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发