摘要
本文把一个求解高维空间数据聚类问题转换为一个超图分割寻优问题,提出一种基于超图模式的高维空间数据聚类方法.该方法不需要减少高维空间数据项的维数,直接用超图模式描述原始数据之间的关系,并能通过选择适当的支持度阈值,有效去除噪声点,保证数据聚类的质量.
In this paper, the problem of solving the data clustering in high dimensional space is formulated as a hypergraph optimal partition problem, and a method is proposed for clustering data in high dimensional space based on hypergraph model. It does not require dimensionality reduction, and it uses the hypergraph model to represent relations among the original data items. By finding an appropriate support threshold, noise data can be filtered out from the clusters very effectively and the quality of the cluseters can be controled.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2002年第2期223-227,共5页
Pattern Recognition and Artificial Intelligence
基金
广东省科委基金(No.2000-J-006-01)
