Welcome to the new version of CaltechAUTHORS. Login is currently restricted to library staff. If you notice any issues, please email coda@library.caltech.edu
Published March 2012 | public
Journal Article

Beyond sparsity: The role of L_1-optimizer in pattern classification

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.

Additional Information

© 2011 Elsevier Ltd. Received 28 September 2010. Received in revised form 23 July 2011. Accepted 22 August 2011. Available online 30 August 2011. The authors would like to thank the anonymous reviewers for their critical and constructive comments and suggestions. This work was partially supported by the Program for New Century Excellent Talents in University of China, the NUST Outstanding Scholar Supporting Program, the National Science Foundation of China under Grant nos. 60973098 and 90820306, National Science Fund for Distinguished Young Scholars, and the Hong Kong RGC General Research Fund.

Additional details

Created:
August 22, 2023
Modified:
October 24, 2023