Global convergence of Oja´s subspace algorithm for principal component extraction

Author

Chen, Tianping ; Hua, Yingbo ; Yan, Wei-Yong

Author_Institution

Dept. of Math., Fudan Univ., Shanghai, China

Volume

9

Issue

1

fYear

1998

fDate

1/1/1998 12:00:00 AM

Firstpage

58

Lastpage

67

Abstract

Oja´s principal subspace algorithm is a well-known and powerful technique for learning and tracking principal information in time series. A thorough investigation of the convergence property of Oja´s algorithm is undertaken in this paper. The asymptotic convergence rates of the algorithm is discovered. The dependence of the algorithm on its initial weight matrix and the singularity of the data covariance matrix is comprehensively addressed

Keywords

convergence of numerical methods; covariance matrices; differential equations; eigenvalues and eigenfunctions; parallel algorithms; statistical analysis; Oja subspace algorithm; asymptotic convergence rates; covariance matrix; differential equations; eigenvalues; feature extraction; global convergence; initial weight matrix; parallel algorithm; principal component analysis; time series; Convergence; Covariance matrix; Data mining; Differential equations; Feature extraction; Information processing; Least squares methods; Principal component analysis; Random processes; Singular value decomposition;

fLanguage

English

Journal_Title

Neural Networks, IEEE Transactions on

Publisher

ieee

ISSN

1045-9227

Type

jour

DOI

10.1109/72.655030

Filename

655030