First-order approximation of Gram-Schmidt orthonormalization beats deflation in coupled PCA learning rules

Möller R (2006)
Neurocomputing 69(13-15): 1582-1590.

Journal Article | Published | English

No fulltext has been uploaded

Abstract
In coupled. learning rules for principal component analysis, eigenvectors and eigenvalues are simultaneously estimated in a coupled system of equations. Coupled single-neuron rules have favorable convergence properties. For the estimation of multiple eigenvectors, orthonormalization methods have to be applied, either full Gram-Schmidt orthonormalization, its first-order approximation as used in Oja's stochastic gradient ascent algorithm, or deflation as in Sanger's generalized Hebbian algorithm. This paper reports the observation that a first-order approximation of Gram-Schmidt orthonormalization is superior to the standard deflation procedure in coupled learning rules. The first-order approximation exhibits a smaller orthonormality error and produces eigenvectors and eigenvalues of better quality. This improvement is essential for applications where multiple principal eigenvectors have to be estimated simultaneously rather than sequentially. Moreover, loss of orthonormality may have an harmful effect on subsequent processing stages, like the computation of distance measures for competition in local PCA methods. (c) 2005 Elsevier B.V. All rights reserved.
Publishing Year
ISSN
PUB-ID

Cite this

Möller R. First-order approximation of Gram-Schmidt orthonormalization beats deflation in coupled PCA learning rules. Neurocomputing. 2006;69(13-15):1582-1590.
Möller, R. (2006). First-order approximation of Gram-Schmidt orthonormalization beats deflation in coupled PCA learning rules. Neurocomputing, 69(13-15), 1582-1590.
Möller, R. (2006). First-order approximation of Gram-Schmidt orthonormalization beats deflation in coupled PCA learning rules. Neurocomputing 69, 1582-1590.
Möller, R., 2006. First-order approximation of Gram-Schmidt orthonormalization beats deflation in coupled PCA learning rules. Neurocomputing, 69(13-15), p 1582-1590.
R. Möller, “First-order approximation of Gram-Schmidt orthonormalization beats deflation in coupled PCA learning rules”, Neurocomputing, vol. 69, 2006, pp. 1582-1590.
Möller, R.: First-order approximation of Gram-Schmidt orthonormalization beats deflation in coupled PCA learning rules. Neurocomputing. 69, 1582-1590 (2006).
Möller, Ralf. “First-order approximation of Gram-Schmidt orthonormalization beats deflation in coupled PCA learning rules”. Neurocomputing 69.13-15 (2006): 1582-1590.
This data publication is cited in the following publications:
This publication cites the following data publications:

Export

0 Marked Publications

Open Data PUB

Web of Science

View record in Web of Science®

Search this title in

Google Scholar