计算机工程与应用 ›› 2007, Vol. 43 ›› Issue (29): 165-167.

• 数据库与信息处理 • 上一篇    下一篇

基于构造型神经网络和商空间粒度的聚类方法

徐 银1,2,周文江1,2,王伦文2    

  1. 1.解放军电子工程学院 研二队,合肥 230037
    2.解放军电子工程学院 309室,合肥 230037
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2007-10-11 发布日期:2007-10-11
  • 通讯作者: 徐 银

Clustering method based on Constructive Neural Networks and quotient space granularity

XU Yin1,2,ZHOU Wen-jiang1,2,WANG Lun-wen2   

  1. 1.The Second Team of Postgraduate Department at Electronic Engineering Institute,Hefei 230037,China
    2.309 Research Division of Electronic Engineering Institute,Hefei 230037,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-10-11 Published:2007-10-11
  • Contact: XU Yin

摘要: 采用构造型神经网络对大规模模式进行聚类,其中利用商空间粒度分析法选择最优粒度聚类。该方法既发挥了构造型神经网络计算复杂度低的优点,又利用了商空间理论选取最优粒度聚类。对大规模复杂数据聚类实验结果表明该方法是实效的。

关键词: 聚类, 粒度, 构造型神经网络, 商空间

Abstract: In this paper,Constructive Neural Networks(i.e.CNN) are used to cluster large-scale patterns,and the optimum granularity is chosen by quotient space granularity analysis method.This method not only makes good use of the characteristic of CNN in reducing the computing complexity,but also takes the advantage of quotient space theory in choosing the optimum granularity.The results of the experiments of clustering large-scale and complicated data show the validity of this method.

Key words: clustering, granularity, constructive neural networks, quotient space