基于区域最近邻生长的层次聚类算法
杨栋1 詹海亮2 苏锦旗3
1.(西京学院 陕西 西安 710123;)2.(西北工业大学 陕西 西安 710072)
摘 要:对于非球形和其它特殊形状的非凸数据集的聚类,基于划分的聚类算法很难取得理想的聚类结果。层次聚类算法根据数据的特征将距离近的数据进行合并,对于球形数据集和其它具有特殊形状的数据集有很好的聚类效果。本文在分析现有层次聚类算法的基础上,根据层次聚类的合并思想和最近邻距离的计算提出了基于区域最近邻生长的层次聚类算法。
关键词:聚类算法;层次聚类算法;区域最近邻生长
中图分类号:TP 文献标识码:A 文章编号:
Based on regional growth and the level of
the nearest neighbor clustering algorithm
YANG Dong1,ZHAN Hai-liang2
1(Xijing College Shaannxi Xi’an 710123, China)
23(Northwestern Polytechnical University, Shaanxi Xi’an 710072, China)
Abstract: For non-spherical shapes, and other special non-convex data sets clustering, clustering algorithm based on division of clustering is difficult to achieve the desired results. Hierarchical clustering algorithm based on the data will be characterized by data from the recent merger of data sets for spherical and other data sets with special shapes have a good clustering efficiency. In this paper, analysis of existing hierarchical clustering algorithm based on the hierarchical clustering of the combined ideas and the nearest neighbor distance calculation is presented based on the growth of the region nearest-neighbor hierarchical clustering algorithm
Key words: Clustering algorithm; hierarchical clustering algorithm; Regional Growth nearest neighbor
1引言
层次聚类是基于分层思想的聚类算法,他的优点是可以在不同维度水平上对数据进行探测,容易实现相似度量或距离度量。但单纯的层次聚类算法终止条件含糊,且执
基于区域最近邻生长的层次聚类算法 来自淘豆网m.daumloan.com转载请标明出处.