Hits:

Indexed by:期刊论文

Date of Publication:2013-01-01

Journal:ICIC Express Letters

Included Journals:EI、Scopus

Volume:7

Issue:8

Page Number:2411-2416

ISSN No.:1881803X

Abstract:In spectral clustering algorithms, informative eigenvectors of a spectral clustering matrix are used to cluster data set. Informative eigenvectors are a group of eigenvectors which include clustering information of the data set. Selecting of informative eigenvectors is one of the key issues in spectral clustering. In this paper, smart selection of informative eigenvectors in spectral clustering was proposed (SSIESC). SSIESC differs from previous approaches in that it can distinguish informative eigenvectors remarkably from uninformative ones. SSIESC was evaluated using synthetic data sets as well as UCI benchmark data sets. Experimental results demonstrated that, the main merit of SSIESC is simple but effective than existing approaches. ? 2013 ICIC International.