首页 | 本学科首页   官方微博 | 高级检索  
     

一种基于网格和密度凝聚点的快速聚类算法
引用本文:陈卓,孟庆春,魏振钢,任丽婕,窦金凤. 一种基于网格和密度凝聚点的快速聚类算法[J]. 哈尔滨工业大学学报, 2005, 37(12): 1654-1657
作者姓名:陈卓  孟庆春  魏振钢  任丽婕  窦金凤
作者单位:中国海洋大学,计算机科学系,山东,青岛,266071;中国海洋大学,计算机科学系,山东,青岛,266071;清华大学,智能技术与系统国家重点实验室,北京,100084
基金项目:国家自然科学基金资助项目(60374031);山东省自然科学基金资助项目(Y2002G18).
摘    要:提出的快速聚类算法通过凝聚点来准确反映数据空间的几何特征,然后采用网格和密度相结合的方法,利用爬山法和连通性原理进行聚类处理,克服了传统网格聚类算法聚类质量降低的缺点.实验结果证明,本算法的聚类效率优于传统爬山法、Clique算法和DBSCAN算法.

关 键 词:聚类  网格  密度  牛顿爬山法  凝聚点
文章编号:0367-6234(2005)12-1654-04
收稿时间:2004-05-12
修稿时间:2004-05-12

A fast clustering algorithm based on grid and density condensation point
CHEN Zhuo,MENG Qing-chun,WEI Zhen-gang,REN Li-jie,DOU Jin-feng. A fast clustering algorithm based on grid and density condensation point[J]. Journal of Harbin Institute of Technology, 2005, 37(12): 1654-1657
Authors:CHEN Zhuo  MENG Qing-chun  WEI Zhen-gang  REN Li-jie  DOU Jin-feng
Abstract:A new kind of clustering algorithm called CGDCP is presented.The creativity of CGDCP is capturing the shape of data space by condensation points,and then using grid-based and density-based clustering methods based on the theories of a climbing hill algorithm and connectivity to deal with the data.CGDCP retains the good features of grid-based and density-based clustering methods and overcomes the traditional shortcomings of the grid-based clustering method's quality debasement resulting from little or no consideration of data distribution when partitioning the grids.Experimental results confirm that the execution efficiency of CGDCP is much better than a traditional climbing hill algorithm and the Clique algorithm.
Keywords:clustering   grid    density    climbing hill algorithm    condensation point
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号