基于leader的k-means改进算法  

Improved k-means Algorithm Based on Leader Algorithm

作　　者：郑富兰[1] 张艳芳[2] 吴瑞[1] 

机构地区：[1]山西师范大学数学与计算机科学学院,山西临汾041000 [2]临汾职业技术学院计算机系,山西临汾041004

出　　处：《山西师范大学学报（自然科学版）》2014年第4期26-29,共4页Journal of Shanxi Normal University(Natural Science Edition)

基　　金：山西省自然科学基金(2008011029-2)

摘　　要：针对传统k-means算法中初始聚类中心选取的随意性对于聚类结果影响较大的问题,提出了基于Leader算法的k-means改进算法——Lk-means算法.该算法有效避免了初始聚类中心选取的边缘化和随意性.实验证明,Lk-means算法的聚类结果更加有效合理.To the problem of the signi ficant effect on the cluster results owing to the randomicity of traditional k-means on determining the initial cluster centers, we propose an improved k-means algorithm, Lk-means, which is based on leader algorithm. The Lk-means algorithm is effective to avoid the marginalization and randomicity emerged in determining the initial cluster centers. Experiment proves that the algorithm makes the cluster results more objective and reasonable.

关 键 词：聚类 Leader算法 K-MEANS算法 改进算法 

分 类 号：G202[文化科学—传播学]