计算机工程与应用 ›› 2011, Vol. 47 ›› Issue (26): 25-26.

• 研究、探讨 • 上一篇    下一篇

动态粗集理论在K-均值聚类中的应用

张 军,黄顺亮   

  1. 山东理工大学 管理学院,山东 淄博 255049
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2011-09-11 发布日期:2011-09-11

Application of dynamic rough sets theory in K-means clustering

ZHANG Jun,HUANG Shunliang   

  1. School of Management,Shandong University of Technology,Zibo,Shangdong 255049,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-09-11 Published:2011-09-11

摘要: 基于动态粗集理论,提出了一种改进的动态粗集K-均值算法。改进后的算法依据数据对象的迁移系数大小,被划分到某一类的膨胀上近似集或膨胀下近似集;在计算类的质心时采用其中数据对象集的迁移系数的平均值作为权值来衡量它对质心的贡献。在UCI机器学习数据库原始数据集及其噪音数据集上的实验结果表明,改进后的动态粗集K-均值算法提高了聚类的准确性,降低了迭代次数。

关键词: 动态粗集, K-均值, 聚类, 迁移系数

Abstract: Based on the theory of dynamic rough set,a dynamic rough K-means clustering algorithm is presented.The new algorithm divides samples into a cluster according to the transition coefficient.When calculating the centroids the average of transition coefficients are used as weight,which represents the contributions of the samples to the cluster.Experiments on UCI data sets and on generated data sets with noise points prove this algorithm can get better clustering accuracy and reduce the iteration times.

Key words: dynamic rough set, K-means, cluster, transition coefficient