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

被引：56

作者：

Chen, TP ^{[1
]}

Hua, YB

Yan, WY

机构：

[1] Fudan Univ, Dept Math, Shanghai 200433, Peoples R China

[2] Univ Melbourne, Dept Elect & Elect Engn, Parkville, Vic 3052, Australia

[3] Nanyang Technol Univ, Sch Elect & Elect Engn, Singapore 639798, Singapore

来源：

IEEE TRANSACTIONS ON NEURAL NETWORKS | 1998年 / 9卷 / 01期

基金：

中国国家自然科学基金; 澳大利亚研究理事会;

关键词：

convergence rate; global convergence; principal components extraction;

D O I：

10.1109/72.655030

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

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 asymptotical 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.

引用

页码：58 / 67

页数：10

共 11 条

[1] NEURAL THEORY OF ASSOCIATION AND CONCEPT-FORMATION [J].

AMARI, SI .

BIOLOGICAL CYBERNETICS, 1977, 26 (03) :175-185

[2] LEARNING IN LINEAR NEURAL NETWORKS - A SURVEY [J].

BALDI, PF ;

HORNIK, K .

IEEE TRANSACTIONS ON NEURAL NETWORKS, 1995, 6 (04) :837-858

[3]

GOLUB GH, 1985, MATRIX COMPUTATION, P271

[4] CONVERGENCE ANALYSIS OF LOCAL FEATURE-EXTRACTION ALGORITHMS [J].

HORNIK, K ;

KUAN, CM .

NEURAL NETWORKS, 1992, 5 (02) :229-240

[5]

Lei Xu, 1991, International Journal of Neural Systems, V2, P169, DOI 10.1142/S0129065791000169

[6] PRINCIPAL COMPONENTS, MINOR COMPONENTS, AND LINEAR NEURAL NETWORKS [J].

OJA, E .

NEURAL NETWORKS, 1992, 5 (06) :927-935

[7] ON STOCHASTIC-APPROXIMATION OF THE EIGENVECTORS AND EIGENVALUES OF THE EXPECTATION OF A RANDOM MATRIX [J].

OJA, E ;

KARHUNEN, J .

JOURNAL OF MATHEMATICAL ANALYSIS AND APPLICATIONS, 1985, 106 (01) :69-84

[8]

Oja E., 1989, International Journal of Neural Systems, V1, P61, DOI 10.1142/S0129065789000475

[9] A SIMPLIFIED NEURON MODEL AS A PRINCIPAL COMPONENT ANALYZER [J].

OJA, E .

JOURNAL OF MATHEMATICAL BIOLOGY, 1982, 15 (03) :267-273

[10] LEAST MEAN-SQUARE ERROR RECONSTRUCTION PRINCIPLE FOR SELF-ORGANIZING NEURAL-NETS [J].

XU, L .

NEURAL NETWORKS, 1993, 6 (05) :627-648

← 1 2 →