| 基于类轮廓层次聚类方法的研究 |
| |
| 引用本文： | 孟海东,唐旋.基于类轮廓层次聚类方法的研究[J].计算机应用与软件,2011,28(11). |
| |
| 作者姓名： | 孟海东 唐旋 |
| |
| 作者单位： | 内蒙古科技大学信息工程学院 内蒙古包头014010 |
| |
| 基金项目： | 国家自然科学基金资助项目(40762003); 教育部“春晖计划”合作研究项目(Z2009-1-01041) |
| |
| 摘 要： | 传统的聚类算法在考虑类与类之间的连通性特征和近似性特征上往往顾此失彼。首先给出类边界点和类轮廓的基本定义以及寻求方法,然后基于类间连通性特征和近似性特征的综合考虑,拟定一些类间相似性度量标准和方法,最后提出一种基于类轮廓的层次聚类算法。该算法能够有效处理任意形状的簇,且能够区分孤立点和噪声数据。通过对图像数据集和Iris标准数据集的聚类分析,验证了该算法的可行性和有效性。
|
| 关 键 词： | 连通性 近似性 类轮廓 层次聚类 |
RESEARCH ON CLASS-PROFILE-BASED HIERARCHICAL CLUSTERING METHOD |
| |
| Meng Haidong,Tang Xuan.RESEARCH ON CLASS-PROFILE-BASED HIERARCHICAL CLUSTERING METHOD[J].Computer Applications and Software,2011,28(11). |
| |
| Authors: | Meng Haidong Tang Xuan |
| |
| Affiliation: | Meng Haidong Tang Xuan (School of Information Engineering,Inner Mongolia University of Science and Technology,Baotou 014010,Inner Mongolia,China) |
| |
| Abstract: | Traditional clustering algorithms are often incapable of roundly considering the connectivity and similarity characteristics among classes.The thesis firstly presents the fundamental definition of class boundary point and class profile;secondly,with comprehensive consideration based on connectivity characteristics and similarity characteristics among classes,defines some standards and methods for inter class similarity measurement;thirdly,proposes a class-profile-based hierarchical clustering algorithm,whic... |
| |
| Keywords: | Connectivity Proximity Class profile Hierarchical clustering |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
