优化的初始中心点选取的K-means聚类算法  

K-means Clustering Algorithm to Optimize the Initial Center Point

作　　者：王金金[1] 王未央[1] 

机构地区：[1]上海海事大学信息工程学院,上海201306

出　　处：《现代计算机（中旬刊）》2015年第7期6-9,共4页Modern Computer

摘　　要：介绍一种可以对初始聚类中心进行优化的算法,改进之处是对孤立点进行特殊处理,降低孤立点敏感的问题,把距离与密度结合,选取最优的初始中心点,从而使聚类的精确度得到提高,并且该算法通过在计算的过程中存储数据对象之间的距离来提高算法的效率。通过对实验结果的分析,得到改进后的聚类算法可以有更好的精确度和更高的算法效率。Describes an algorithm which initial cluster centers can be optimized. The improvement is to isolate point for special treatment, reduce outlier sensitive issue, combines the distance and density to select the appropriate initial focal point, so that improves the clustering accu- racy, in order to improve the efficiency of the algorithm, the algorithm in the process of calculating the distance between the stored data objects. The experimental results prove that the improved clustering algorithm can achieve better results and higher efficiency of the algo- rithm.

关 键 词：K—means算法 聚类中心 孤立点 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]