摘要
针对初始聚类中心对传统K-means算法的聚类结果有较大影响的问题,提出一种依据样本点类内距离动态调整中心点类间距离的初始聚类中心选取方法,由此得到的初始聚类中心点尽可能分散且具代表性,能有效避免K-means算法陷入局部最优。通过UCI数据集上的数据对改进算法进行实验,结果表明改进的算法提高了聚类的准确性。
There are great impacts on traditional K-means algorithm results of clustering for initial cluster centers. A new im- proved K-means algorithm is proposed. A new method for selecting initial cluster centers according to the inner class distance of samples which dynamically adjust the distance between clustering. It not only can nake the cluster centers as dispersed as possible and highly representative ,but can avoid K-means algorithm into local optimum effectively. The improved algorithm is done experi- ments on data of UCI data set, the results show that improved algorithm can improve the clustering accuracy.
出处
《微型机与应用》
2012年第20期74-76,共3页
Microcomputer & Its Applications
基金
福建省自然科学基金(2012J01263)
福建省科技计划重点项目(2011Y0040)
