计算机应用 ›› 2013, Vol. 33 ›› Issue (04): 991-993.DOI: 10.3724/SP.J.1087.2013.00991

• 人工智能 • 上一篇    下一篇

模糊C均值与支持向量机相结合的增强聚类算法

胡磊,牛秦洲,陈艳   

  1. 桂林理工大学 信息科学与工程学院,广西 桂林 541004
  • 收稿日期:2012-10-23 修回日期:2012-11-18 出版日期:2013-04-01 发布日期:2013-04-23
  • 通讯作者: 胡磊
  • 作者简介:胡磊(1986-),男,山西太原人,硕士研究生,CCF会员,主要研究方向:制造业信息化、人工智能;牛秦洲(1956-),男,陕西西安人,教授,博士,主要研究方向:制造业信息化、人工智能;陈艳(1987-),女,广西桂林人,硕士研究生,主要研究方向:制造业信息化。
  • 基金资助:

    广西科技攻关项目(桂科攻[10100002-2]号);广西研究生教育创新计划项目(2011105960812M23);桂林市科学研究与技术开发计划项目(桂林市201101102-2)

Enhanced clustering algorithm based on fuzzy C-means and support vector machine

HU Lei,NIU Qinzhou,CHEN Yan   

  1. College of Information Science and Engineering, Guilin University of Technology, Guilin Guangxi 541004, China
  • Received:2012-10-23 Revised:2012-11-18 Online:2013-04-01 Published:2013-04-23
  • Contact: HU Lei

摘要: 针对传统重复聚类算法精度不高、消耗资源较大的缺点,提出了一种模糊C均值(FCM)与支持向量机(SVM)相结合的增强聚类算法。该算法思路是先将实例数据集利用FCM粗分为C类,然后使用SVM再对每一类进行细化分类,实现中提出了基于完全二叉树的决策级联式SVM模型,以便达到增强聚类的目的。针对使用FCM迭代聚类的过程中有可能会出现新的特征使原有的聚类失去平衡性的问题,提出了使用划分的思想对数据集进行预处理来消除这种不利影响。利用鸢尾属植物真实数据集对相关算法进行实验对比分析,结果表明该算法能够克服精度低的缺点,并节约了系统资源,可以提高聚类的质量。

关键词: 模糊C均值, 支持向量机, 增强聚类, 完全二叉树, 量化指标评价

Abstract: To improve the accuracy and efficiency of clustering algorithm, this paper proposed an enhanced algorithm based on Fuzzy C-Means (FCM) and Support Vector Machine (SVM). The sets of data were clustered into c kinds by FCM, and then they were classified by SVM in detail. The cascade SVM model based on fully binary decision tree was constructed, so as to enhance clustering. In order to solve the problem of losing balance in making new features, the idea of using division in a set of data to eliminate the bad effect was put forward. Some correlation algorithms were compared on Iris data set. The experimental results show that the algorithm can improve the precision, save the system resources and enhance the efficiency of clustering.

Key words: Fuzzy C-Means(FCM), Support Vector Machine (SVM), enhanced clustering, fully binary tree, quantitative index evaluation

中图分类号: