Adaptive Dimension Reduction for Clustering High Dimensional Data.

Ding, C. ; He, X. ; Zha, H. ; Simon, H. D.

Technical Information Center Oak Ridge Tennessee

关键词: Algorithms;    Clustering;    Dimensions;    Iterations;    Attributes;   

RP-ID  :  DE2003807420

学科分类：工程和技术（综合）

美国|英语

来源: National Technical Reports Library

【 摘 要 】

It is well-known that for high dimensional data clustering, standard algorithms such as EM and the K-means are often trapped in local minimum. Many initialization methods were proposed to tackle this problem, but with only limited success. In this paper we propose a new approach to resolve this problem by repeated dimension reductions such that K-means or EM are performed only in very low dimensions. Cluster membership is utilized as a bridge between the reduced dimensional subspace and the original space, providing flexibility and ease of implementation. Clustering analysis performed on highly overlapped Gaussians, DNA gene expression profiles and internet newsgroups demonstrate the effectiveness of the proposed algorithm.