设为首页 联系我们 加入收藏

当前位置: 网站首页 期刊分类目录 2019第5期 正文

基于KNN算法的改进K-means算法

作者:时间:2019-10-25点击数:

PDF全文下载:  201905016.pdf

 

文章编号: 16726987201905010705 DOI 10.16351/j.16726987.2019.05.016

 

徐文进, 管克航, 寻晴晴, 许瑶, 解钦

(青岛科技大学 信息科学技术学院,山东 青岛 266061)

 

摘要: 针对于K-means算法的缺点做出了一些改进,提出了一种基于KNN算法改进K-means的算法。改进后的算法解决了K-means算法K值无法确定和数据分类中的不强、易受异常数据干扰的缺点,提高了算法的聚类效果以及削弱初始聚类中心选择的随机性对于聚类结果易陷入局部最优的影响。实验表明,改进后的算法不仅解决了传统算法确定K值的问题,而且聚类结果稳定且聚类效果良好。

关键词: 数据中心点; K-means; 局部最优; KNN算法

 

中图分类号: TP 181文献标志码: A

引用格式: 徐文进, 管克航, 寻晴晴, . 基于KNN算法的改进K-means算法\[J\]. 青岛科技大学学报(自然科学版), 2019 405): 107111.

 

XU Wenjin, GUAN Kehang, XUN Qingqing, et al. Improving K-means algorithm based on KNN algorithm\[J\]. Journal of Qingdao University of Science and TechnologyNatural Science Edition), 2019, 40(5) 107111.

Improving Kmeans Algorithm Based on KNN Algorithm

 

XU Wenjin, GUAN Kehang, XUN Qingqing, XU Yao, XIE Qin

(College of Information Science and Technology, Qingdao University of Science and Technology, Qingdao 266061,China)

 

Abstract: In this paper, some improvements are made to the shortcomings of K-means algorithm, and an algorithm based on KNN algorithm to improve K-means is proposed. The improved algorithm solves the shortcomings of the Kmeans algorithm K value can not be determined and the data classification is not strong, susceptible to abnormal data interference, improve the clustering effect of the algorithm and weaken the initial cluster center selection random for the clustering results easy to fall into the local optimal impact. Experiments show that the improved algorithm not only solves the problem of determining the K value by the traditional algorithm, but also has stable clustering results and good clustering effect.

Key words: data center point; K-means; local optimum; KNN algorithm

 

收稿日期:  20181123

基金项目: 山东省重点研发计划项目(2018GGX105005.

作者简介: 徐文进(1977—),男,副教授.

 

Copyright © 2011-2017 青岛科技大学学报 (自然科学版)