WSEAS Transactions on Mathematics
Print ISSN: 1109-2769, E-ISSN: 2224-2880
Volume 13, 2014
Weighted Generalized LDA for Undersampled Problems
Authors: , ,
Abstract: Linear discriminant analysis (LDA) is a classical approach for dimensionality reduction. It aims to max imize between-class scatter and minimize within-class scatter, thus maximize the class discriminant. However, for undersampled problems where the data dimensionality is larger than the sample size, all scatter matrices are singular and the classical LDA encounters computational difficulty. Recently, many LDA extensions have been developed to overcome the singularity, such as, Pseudo-inverse LDA (PILDA), LDA based on generalized singular value decomposition (LDA/GSVD), null space LDA (NLDA) and range space LDA (RSLDA). Moreover, they endure the Fisher criterion that is nonoptimal with respect to classification rate. To remedy this problem, weighted schemes are presented for several LDA extensions in this paper and called them weighted generalized LDA algo rithms. Experiments on Yale face database and AT&T face database are performed to test and evaluate effective of the proposed algorithms and affect of weighting functions.
Search Articles
Keywords: Feature extraction, dimensionality reduction, undersampled problem, weighting function, misclassification rate
Pages: 694-703
WSEAS Transactions on Mathematics, ISSN / E-ISSN: 1109-2769 / 2224-2880, Volume 13, 2014, Art. #68