Influence of the Learning Gain on the Dynamics of Oja’s Neural Network

  • Pedro J. Zufiria
Conference paper


In this paper, the dynamical behavior of Oja’s neural network [7] is analyzed. Oja’s net has been traditionally studied in the continuous-time context via some simplification procedures, some of them concerning the asymptotic behavior of the learning gain. The contribution of the paper is the study of a deterministic discrete-time (DDT) version, preserving the discrete-time form of the original network and allowing a more realistic behavior of the learning gain. As a consequence, the discrete-time nature of the new model leads to results which are drastically different to the ones known for the continuous-time formulation. Simulation examples support the presented results.


Invariant Subspace Chaotic Behavior Invariant Manifold Learning Gain Stable Fixed Point 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    F. T. Arecchi, R. Badii and A. Politi, Low-frequency phenomena in dynamical systems with many attractors Phys. Rev. A, Vol. 29, NO. 2, pp. 1006–1009, 1984.CrossRefGoogle Scholar
  2. [2]
    J. A. Berzal, P. J. Zufiria and L. Rodriguez, Implementing the Karhunen-Loeve transform via improved neural networks. Solving Engineering Problems with Neural Networks. International Conference on Engineering Applications of Neural Networks, EANN′96. pp. 375–378, London 1996. ISBN 952-90-7517-0.Google Scholar
  3. [3]
    H. J. Kushner & D. S. Clark, Stochastic Approximation Methods for Constrained and Unconstrained Systems New York, Springer-Verlag, 1978.CrossRefGoogle Scholar
  4. [4]
    S. Haykin, Neural Networks. A comprehensive foundation Macmillan Publishing Company, 1994.Google Scholar
  5. [5]
    R.-W. Liu, Y.-F. Huang, and X.-T. Ling A Novel Approach to the Convergence of Neural Networks for Signal Processing. IEEE Transactions on Circuits and Systems, Vol. 42, NO. 3, 187–188, 1995.MATHCrossRefGoogle Scholar
  6. [6]
    L. Ljung, Analysis of recursive stochastic algorithms IEEE Transactions on Automatic Control AC-22, 551–575, 1977.MathSciNetCrossRefGoogle Scholar
  7. [7]
    E. Oja, A Simplified Neuron Model as a Principal Component Analyzer Journal of Mathematical Biology, 15, 267–273, 1982.MathSciNetMATHCrossRefGoogle Scholar
  8. [8]
    F. Peper and H. Noda, A Symmetric Linear Neural Network That Learns Principal Components and Their Variances IEEE Transactions on Neural Networks Vol. 7, NO. 4, 1042–1047, 1996.CrossRefGoogle Scholar
  9. [9]
    M. D. Plumbley, Lyapunov Functions for Convergence of Principal Component Algorithms Neural Networks, Vol. 8, NO. 1, 11–23, 1995.CrossRefGoogle Scholar
  10. [10]
    H. Robbins and S. Monro, A Stochastic Approximation Method Ann. Math. Stat. 22, 400–407, 1951.MathSciNetMATHCrossRefGoogle Scholar
  11. [11]
    T. D. Sanger, Optimal unsupervised learning in a single-layer linear feedforward neural network Neural Networks, vol. 2, pp. 459–473, 1989.CrossRefGoogle Scholar
  12. [12]
    J. Testa and G. A. Help, Study of a one-dimensional map with multiple basins Phys. Rev. A., Vol. 28, NO. 3, pp. 3085–3089, 1983.CrossRefGoogle Scholar
  13. [13]
    W.-Y. Yan, U. Helmke, and J.B. Moore, Global Analysis of Oja’s Flow for Neural Networks IEEE Transactions on Neural Networks Vol. 5, NO. 5, 674–683, 1994.CrossRefGoogle Scholar
  14. [14]
    Q. Zhang, and Z. Bao, Dynamical System for Computing the Eigenvectors Associated with the Largest Eigenvalue of a Positive Definite Matrix IEEE Transactions on Neural Networks Vol. 6, NO. 3, 790–791, 1995.MathSciNetCrossRefGoogle Scholar
  15. [15]
    Q. Zhang, and Y. W. Leung, Energy function for One-Unit Oja Algorithm IEEE Transactions on Neural Networks Vol. 6, NO. 5, 1291–1293, 1995.CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Wien 1999

Authors and Affiliations

  • Pedro J. Zufiria
    • 1
  1. 1.Grupo de Redes Neuronales Depto. Matemática Aplicada a las Tecnologías de la Informatión Escuela Técnica Superior de Ingenieros de TelecomunicaciónUniversidad Politécnica de Madrid Ciudad Universitaria s/n.MadridSpain

Personalised recommendations