Advertisement

A new approach for extracting rules from a trained neural network

  • Castellanos A. L. 
  • Castellanos J. 
  • Manrique D. 
  • Martinez A. 
Posters (Extended Abstracts)
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1323)

Abstract

Artificial Neural Networks perform adaptive learning. This advantage can be used to complete and improve the knowledge acquisition in knowledge engineering by rule extraction from a trained neural network. This paper proposes a new rule extraction method based on MACIE algorithm, which has been improved so that it can be used in neural networks with continuous inputs and outputs, obtaining a global and continuous set of production rules in a very efficient way. An application example to obtain the average load demand of a power plant is also shown.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. [1]
    Carpintero A., Castellanos J., Mingo L.F., Rios J., “Short-Term Load Demand with a Mixed Neural Network System” Proceedings of ICICS'96, Tehran pp. 60–63, 1996.Google Scholar
  2. [2]
    Castellanos J., Pazos A., Rios J., Zafra J.L., “Sensitivity Analysis on Neural Networks for Metereological Variable Forecasting” Neural Networks for signal processing IV. Proceedings of the 1994 IEEE workshop, New York pp. 587–595, 1994.Google Scholar
  3. [3]
    Craven M. and Shavlik J., “Learning Symbolic Rules using Artificial Neural Networks”, Proceedings of the 10th International Conference on Machine Learning, 1993, pp.73–80.Google Scholar
  4. [4]
    Síma J., “Neural Expert Systems”, Neural Networks, Vol 8, No. 2, pp. 261–271, 1995.Google Scholar
  5. [5]
    Gallant S. I. “Neural Network Learning and Expert Systems”, MIT Press, Massachusetts, 1993.Google Scholar
  6. [6]
    Fu L.M. “Rule Generation from Neural Networks”. IEEE Transactions on Systems, Man, and Cybernetics, 24(8), 1994, pp. 1114–1124.Google Scholar
  7. [7]
    Zurada J.M., Malinowski A., Cloete I., “Sesitivity Analysis for Minimization of Input Data Dimension for Feedforward Neural Network”, IEEE International Symposium on Circuits and Systems, London, 1994.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1997

Authors and Affiliations

  • Castellanos A. L. 
    • 1
  • Castellanos J. 
    • 2
  • Manrique D. 
    • 1
  • Martinez A. 
    • 2
  1. 1.Departamento de Inteligencia Artificial Facultad de InformaticaUniversidad Politecnica de MadridBoadilla de Monte - MadridSpain
  2. 2.Escuela Universitaria de Ingenieros Técnicos ForestalesUniversidad Politecnica de MadridMadridSpain

Personalised recommendations