Random subspace based semi-supervised feature selection

Author

Ren, Ya-zhou ; Zhang, Guo-ji ; Yu, Guo-xian

Author_Institution

Sch. of Comput. Sci. & Eng., South China Univ. of Technol., Guangzhou, China

Volume

1

fYear

2011

fDate

10-13 July 2011

Firstpage

113

Lastpage

118

Abstract

Feature selection is important in data mining, especially in mining high-dimensional data. In this paper, a random subspace based semi-supervised feature selection (RSSSFS) method with pairwise constraints is proposed. Firstly, several graphs are constructed by different random subspaces of samples, and then RSSSFS combines these graphs into a mixture graph on which RSSSFS does feature selection. The RSSSFS score reflects both the locality preserving power and pairwise constraints. We compare RSSSFS with Laplacian Score and Constraint Score algorithms. Experimental results on several UCI data sets demonstrate its effectiveness.

Keywords

constraint handling; data mining; graph theory; Laplacian score; UCI data sets; constraint score algorithms; high dimensional data mining; locality preserving power; mixture graph; pairwise constraints; random subspace based semisupervised feature selection; Accuracy; Data mining; Ionosphere; Iris; Laplace equations; Machine learning; Sonar; Feature selection; Mixture graph; Pairwise constraints; Random subspaces;

fLanguage

English

Publisher

ieee

Conference_Titel

Machine Learning and Cybernetics (ICMLC), 2011 International Conference on

Conference_Location

Guilin

ISSN

2160-133X

Print_ISBN

978-1-4577-0305-8

Type

conf

DOI

10.1109/ICMLC.2011.6016706

Filename

6016706