Canonical Genetic Learning of RBF Networks Is Faster

  • R. Neruda
Conference paper


We extend our previous theoretical results concerning functional equivalence of Gaussian RBF networks and test the proposed canonical genetic learning algorithm on two problems. In our experiments, canonical learning achieved the same error threshold about two times faster in comparison to standard GA.


Radial Basis Function Network Hide Unit Error Threshold Previous Theoretical Result Threshold canonIcal 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    F. Albertini and E.D. Sontag. For neural networks, function determines form. Neural Networks, 6:975–990, 1993.CrossRefGoogle Scholar
  2. [2]
    R. Hecht-Nielsen. On the algebraic structure of feedforward network weight spaces. In Advanced Neural Computers, pages 129–135. Elsevier, 1990.Google Scholar
  3. [3]
    K. Hlaváčková and R. Neruda. Radial basis function networks. Neural Network World, (1):93–101, 1993.Google Scholar
  4. [4]
    V. Kurkova and P. Kainen. Functionally equivalent feedforward networks. Neural Computation, 6:543–558, 1993.MathSciNetCrossRefGoogle Scholar
  5. [5]
    V. Kurkova and R. Neruda. Uniqueness of the functional representations for the gaussian basis functions. In Proceedings of the ICANN’94, pages 474–477, London, 1994. Springer.Google Scholar
  6. [6]
    R. Neruda. Functional equivalence and genetic learning of RBF networks. In D.W. Pearson, N. C. Steele, R.F. Albrecht (editors), Artificial Neural Networks and Genetic Algorithms, pages 53–56, Wien New York, 1995. Springer Verlag.Google Scholar
  7. [7]
    H.J. Sussmann. Uniqueness of the weights for minimal feedforward nets with a given input-output map. Neural Networks, 5(4):589–594, 1992.CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Wien 1998

Authors and Affiliations

  • R. Neruda
    • 1
  1. 1.Institute of Computer ScienceCzech Academy of SciencesPrague 8Czech Republic

Personalised recommendations