Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/12908
Title: Beyond sparsity : the role of L 1-optimizer in pattern classification
Authors: Yang, J
Zhang, L 
Xu, Y
Yang, JY
Keywords: Classifier
Feature extraction
Pattern classification
Sparse representation
Issue Date: 2012
Publisher: Elsevier
Source: Pattern recognition, 2012, v. 45, no. 3, p. 1104-1118 How to cite?
Journal: Pattern recognition 
Abstract: The newly-emerging sparse representation-based classifier (SRC) shows great potential for pattern classification but lacks theoretical justification. This paper gives an insight into SRC and seeks reasonable supports for its effectiveness. SRC uses L 1-optimizer instead of L 0- optimizer on account of computational convenience and efficiency. We re-examine the role of L 1-optimizer and find that for pattern recognition tasks, L 1-optimizer provides more classification meaningful information than L 0-optimizer does. L 0-optimizer can achieve sparsity only, whereas L 1-optimizer can achieve closeness as well as sparsity. Sparsity determines a small number of nonzero representation coefficients, while closeness makes the nonzero representation coefficients concentrate on the training samples with the same class label as the given test sample. Thus, it is closeness that guarantees the effectiveness of the L 1-optimizer based SRC. Based on the closeness prior, we further propose two kinds of class L 1-optimizer classifiers (CL 1C), the closeness rule based CL 1C (C-CL 1C) and its improved version: the Lasso rule based CL 1C (L-CL 1C). The proposed classifiers are evaluated on five databases and the experimental results demonstrate advantages of the proposed classifiers over SRC in classification performance and computational efficiency for large sample size problems.
URI: http://hdl.handle.net/10397/12908
ISSN: 0031-3203
EISSN: 1873-5142
DOI: 10.1016/j.patcog.2011.08.022
Appears in Collections:Journal/Magazine Article

Access
View full-text via PolyU eLinks SFX Query
Show full item record

SCOPUSTM   
Citations

88
Last Week
0
Last month
2
Citations as of Aug 17, 2017

WEB OF SCIENCETM
Citations

84
Last Week
1
Last month
1
Citations as of Aug 21, 2017

Page view(s)

58
Last Week
3
Last month
Checked on Aug 20, 2017

Google ScholarTM

Check

Altmetric



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.