摘要
提出一种基于一个调整后的准则函数最小化来确定聚类数的算法,算法中把邻近值看成一个尺度参数,通过改变尺度参数的范围值获得聚类中心的数量。最后通过对UCI机器学习数据库中的几个数据库进行实验,证实此方法是比较有效的。
This paper proposes an algorithm to determine the number of clusters based on the minimization of a regularized cost function.This algorithm proposes to use the neighborhood as a scale parameter and obtain the number of Cluster centers at varying values of range of the scale parameter.Some databases which are selected from UCI repository of machine learning database,are used to prove the validity of this algorithm.
出处
《计算机与数字工程》
2007年第2期42-44,共3页
Computer & Digital Engineering
关键词
聚类
分割
等级空间
clustering,partitioning,scale-space
