INDIGO Home University of Illinois at Urbana-Champaign logo uic building uic pavilion uic student center

Sparse supervised dimension reduction in high dimensional classification

Show full item record

Bookmark or cite this item: http://hdl.handle.net/10027/7683

Files in this item

File Description Format
PDF http___projecte ... dfview_1&handle=euclid.pdf (257KB) (no description provided) PDF
Title: Sparse supervised dimension reduction in high dimensional classification
Author(s): Wang, Junhui; Wang, Lifeng
Subject(s): dimension subspace
Abstract: Supervised dimension reduction has proven effective in analyzing data with complex structure. The primary goal is to seek the reduced subspace of minimal dimension which is sufficient for summarizing the data structure of interest. This paper investigates the supervised dimension reduction in high dimensional classification context, and proposes a novel method for estimating the dimension reduction subspace while retaining the ideal classification boundary based on the original dataset. The proposed method combines the techniques of margin based classification and shrinkage estimation, and can estimate the dimension and the directions of the reduced subspace simultaneously. Both theoretical and numerical results indicate that the proposed method is highly competitive against its competitors, especially when the dimension of the covariates exceeds the sample size.
Issue Date: 2010
Publisher: Institute of Mathematical Statistics
Citation Info: Wang, J. H. & Wang, L. F. 2010. Sparse supervised dimension reduction in high dimensional classification. Electronic Journal of Statistics, 4: 914-931. DOI: 10.1214/10-EJS572.
Type: Article
Description: The original source for this publication is at the Institute of Mathematical Statistics; DOI: 10.1214/10-EJS572
URI: http://hdl.handle.net/10027/7683
ISSN: 1935-7524
Sponsor: The research of the second author was supported by a NSF grant DMS-1007634.
Date Available in INDIGO: 2011-05-26
 

This item appears in the following Collection(s)

Show full item record

Statistics

Country Code Views
United States of America 285
China 242
Russian Federation 37
Germany 12
United Kingdom 10

Browse

My Account

Information

Access Key