Advertisement

Neural Processing Letters

, Volume 20, Issue 2, pp 105–124 | Cite as

Lazy Learning in Radial Basis Neural Networks: A Way of Achieving More Accurate Models

  • José M. Valls
  • Inés M. Galván
  • Pedro Isasi
Article

Abstract

Radial Basis Neural Networks have been successfully used in a large number of applications having in its rapid convergence time one of its most important advantages. However, the level of generalization is usually poor and very dependent on the quality of the training data because some of the training patterns can be redundant or irrelevant. In this paper, we present a learning method that automatically selects the training patterns more appropriate to the new sample to be approximated. This training method follows a lazy learning strategy, in the sense that it builds approximations centered around the novel sample. The proposed method has been applied to three different domains

an artificial regression problem and two time series prediction problems. Results have been compared to standard training method using the complete training data set and the new method shows better generalization abilities.

improving generalization ability kernel functions K-means algorithm lazy learning radial basis neural networks 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Moody, J. E. and Darkn, C.: Fast learning in networks of locally tuned processing units. Neural Computation, 1 (1989), 281-294.Google Scholar
  2. 2.
    Poggio, T. and Girosi, F.: Networks for approximation and learning. Proc. IEEE, 78 (1990), 1481-1497.Google Scholar
  3. 3.
    Ghosh, J. and Nag, A.: An Overview of Radial Basis Function Networks. R. J. Howlett and L.C. Jain (Eds). Physica Verlag, (2000).Google Scholar
  4. 4.
    Broomhead, D. S. and Lowe, D.: Multivariable functional interpolation and adaptative networks. Complex Systems, 2 (1988), 321-355.Google Scholar
  5. 5.
    Powell, M.: The theory of radial basis function approximation in 1990. Advances in Numerical Analysis, 3 (1992), 105-210.Google Scholar
  6. 6.
    Park, I. J. and Sandberg, W.: Universal approximation and radial-basis-function networks. Neural Computation, 5 (1993), 305-316.Google Scholar
  7. 7.
    Abu-Mostafa, Y. S.: The vapnik-chervonenkis dimension: information versus complexity in learning. Neural Computation, 1 (1989), 312-317.Google Scholar
  8. 8.
    Cohn, D. Atlas, L. and Ladner.R.: Improving generalization with active learning. machine learning. Machine Learning, 15 (1994) 201-221.Google Scholar
  9. 9.
    Atkenson, C.G. Moore, A.W. and Schaal, S.: Locally weighted learning. Artificial Intelligence Review, 11 (1997) 11-73.Google Scholar
  10. 10.
    Wettschereck, D. and Dietterich, T.: Improving the perfomance of radial basis function networks by learning center locations. Advances in Neural Information Processing Systems, 4 (1992), 1133-1140.Google Scholar
  11. 11.
    Leonardis, A. and Bischof, H.: An efficient mdl-based construction of rbf networks. Neural Networks, 11 (1998), 963-973.Google Scholar
  12. 12.
    Orr. M. J. L.: Introduction to radial basis neural networks. Technical Report. Centre for Cognitive Science, University of Edinburgh, (1996).Google Scholar
  13. 13.
    Yingwei, L. Sundararajan, N. and Saratchandran, P.: A sequential learning scheme for function approximation using minimal radial basis function neural networks. Neural Computation, 9 (1997), 461-478.Google Scholar
  14. 14.
    Mackey, M. C. and Glass, L. Oscillation and chaos in physiological control systems. Science, 197 (1977) 287-289.Google Scholar
  15. 15.
    Platt, J.: A resource-allocating network for function interpolation. Neural Computation, 3 (1991), 213-225.Google Scholar
  16. 16.
    Whitehead, B. A. and Choate, T. D.: Cooperative-competitive genetic evolution of radial basis function centeres and widths for time series prediction. IEEE Transactions on Neural Networks, 5 (1995), 15-23.Google Scholar
  17. 17.
    Moretti, E. and Tomasin, A.: Un contributo matematico all-elaborazione previsionale dei dati di marea a Venecia. Boll. Ocean. Teor. Appl., 1 (1984), 45-61.Google Scholar
  18. 18.
    Michelato, A. Mosetti, R. and Viezzoli. D.: Statistical forecasting of strong surges and aplication to the lagoon of Venice. Boll. Ocean. Teor. Appl., 1 (1983), 67-83.Google Scholar
  19. 19.
    Tomasin, A.: A computer simulation of the Adriatic Sea for the study of its dynamics and for the forecasting of floods in the town of Venice. Comp. Phys. Comm., 5 (1973), 51.Google Scholar
  20. 20.
    Vittori, G.: On the chaotic features of tide elevation in the lagoon Venice. Proc. of the ICCE-92, 23rd International Conference on Coastal Engineering, pages (1992), 4-9.Google Scholar
  21. 21.
    Zaldı´ var, J. M. Gutrrez, E. Galva´ n, I. M. Strozzi, F. and Tomasin, A.: Forecasting high waters at Venice Lagoon using chaotic time series analysis and nonlinear neural networks. Journal of Hydroinformatics, 2 (2000), 61-84.Google Scholar

Copyright information

© Kluwer Academic Publishers 2004

Authors and Affiliations

  • José M. Valls
    • 1
  • Inés M. Galván
    • 1
  • Pedro Isasi
    • 1
  1. 1.Departamento de InformáticaUniversidad Carlos III de MadridLeganésSpain

Personalised recommendations