| 基于三种近邻网络的聚类算法研究 |
| |
| 引用本文： | 马闯,吴涛,段梦雅.基于三种近邻网络的聚类算法研究[J].佳木斯大学学报,2014(5):779-782. |
| |
| 作者姓名： | 马闯 吴涛 段梦雅 |
| |
| 作者单位： | 安徽大学数学科学学院,安徽 合肥,230601 |
| |
| 基金项目： | 国家自然基金，安徽省高等学校省级自然科学研究重点项目(KJ2013A033)． |
| |
| 摘 要： | 根据K近邻、共享K近邻和互K近邻三种近邻算法的思想分别构造复杂网络,然后通过复杂网络的社团发现算法来实现对样本的聚类.最后,将三种方法分别在人工构造的非凸类簇数据集和UCI数据集上进行仿真实验,结果表明三种方法都是可行的,且互K近邻网络聚类方法还具有识别一定数量孤立点功能.
|
| 关 键 词： | K近邻 共享K近邻 互K近邻 社团发现 聚类 |
Research of Clustering Algorithm Based on Three Nearest Neighbor Networks |
| |
| MA Chuang,WU Tao,DUAN Meng-ya.Research of Clustering Algorithm Based on Three Nearest Neighbor Networks[J].Journal of Jiamusi University(Natural Science Edition),2014(5):779-782. |
| |
| Authors: | MA Chuang WU Tao DUAN Meng-ya |
| |
| Institution: | ( Department of Mathemalleal Sciences, Anhui University, Hefei 230601, China) |
| |
| Abstract: | This paper first built complex networks based on three algorithms of nearest neighbor .The al-gorithms were K-nearest neighbor , shared K-nearest neighbors and mutual K -nearest neighbor respectively . Then algorithms of community detection was used to achieve samples clustering .Finally, simulations on datasets of non-convex shape clusters and UCI were carried out based on the above three methods .The experimental re-sults show that three methods are all feasible , and the method of Mutual K -nearest neighbor can recognize some isolated points . |
| |
| Keywords: | K-nearest neighbor shared K-nearest neighbors mutual K-nearest neighbor communi-ty detection clustering |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
