| 未确知均值聚类 |
| |
| 引用本文： | 庞彦军,刘立民,刘开第.未确知均值聚类[J].河北煤炭建筑工程学院学报,2010(4):98-100. |
| |
| 作者姓名： | 庞彦军 刘立民 刘开第 |
| |
| 作者单位： | 河北工程大学理学院,河北邯郸056038 |
| |
| 基金项目： | 国家自然科学基金资助（60874116;60940036）; 河北省自然科学基金资助（F2009000857） |
| |
| 摘 要： | 利用未确知系统理论分析特征对样本分类所作贡献,定义特征的分类权重,并作为启发性知识用于确定样本与各类间的加权距离及样本属于各类的隶属度,建立未确知均值聚类算法。IRIS数据检验表明,未确知均值聚类算法误判样本数少、收敛速度快、鲁棒性好,是一种实用、有效的无监督聚类算法。
|
| 关 键 词： | 均值聚类 分类权重 未确知系统 隶属度 |
Uncertain means clustering |
| |
| Authors: | PANG Yan-jun LIU Li-min LIU Kai-di |
| |
| Affiliation: | (School of Science,Hebei University of Engineering,Hebei Handan,056038,China) |
| |
| Abstract: | In this paper,we firstly defined the classified weight according to the contribution of feature to sample which was used to determined the membership degree of sample to each class and the weighed distance between sample and each class,and then we proposed a new algorithm—the uncertain means clustering.The data of IRIS indicates that the algorithm possesses the better convergence,better robustness and it is an unsupervised clustering algorithm. |
| |
| Keywords: | means clustering classified weight unascertained system membership degree |
| 本文献已被 维普 等数据库收录！ |
|
