摘要
为了解决初始聚类中心的选择、簇个数的确定以及孤立点的避免等问题,提出了一种改进的全局K′-means算法.改进的算法不仅能够利用辅助聚类函数来计算初始点,而且能够利用目标函数在没有预定义聚类个数的前提下,找到实际的聚类中心个数,同时避免了孤立点问题.将改进的算法应用到实际数据集的分类中,并与改进的全局K-means算法以及K′-means算法进行了比较,实验结果证明所提出的算法能获得更好的聚类结果.
In order to solve the problems of the initialization of clustering centers,determination of the number of clustering centers,and avoidance of dead-unit and so on,a modified global E-means algorithm（MGK＇M） is proposed.The improved algorithm can be used not only to calculate its starting point by the auxiliary cluster function,but also to find the actual number of clusters by using the cost-function without preset the number of clusters.At the same time,it can avoid the dead-unit problem.The improved algorithm is used for clustering of actual data sets.Experiment results demonstrate that the proposed algorithm can get better clustering results compared with the modified global K-means algorithm and K-means algorithm.
出处
《信息与控制》
CSCD
北大核心
2011年第1期100-104,共5页
Information and Control
基金
国家863计划资助项目(2008AA04Z131)
国家973计划资助项目(2007CB714300)
北京市优秀人才培养资助项目(2009D013000000003)
