Using evolution to improve neural network learning: pitfalls and solutions (original) (raw)

Abstract

Autonomous neural network systems typically require fast learning and good generalization performance, and there is potentially a trade-off between the two. The use of evolutionary techniques to improve the learning abilities of neural network systems is now widespread. However, there are a range of different evolutionary approaches that could be applied, and no systematic investigation has been carried out to find which work best. In this paper, such an investigation is presented, and it is shown that a range of evolutionary techniques can generate high performance networks, but they often lead to unwanted side effects, such as occasional instances of very poor performance. The nature of these problems are explored further, and it is shown how the evolution of age dependent plasticities and/or the use of ensemble techniques can alleviate them. A range of techniques are thus identified, with differing properties, that can be matched to the specific requirements of each application.

Access this article

Log in via an institution

Subscribe and save

Buy Now

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

Neuroevolution

Chapter © 2017

References

  1. Bishop CM (1995) Neural networks for pattern recognition. Oxford University Press, Oxford
    Google Scholar
  2. Yao X (1999) Evolving artificial neural networks. Proceedings of the IEEE 87:1423–1447
  3. Cantû-Paz E, Kamath C (2005) An empirical comparison of combinations of evolutionary algorithms and neural networks for classification problems. IEEE Trans Syst Man Cybernetics Part B 35:915–927
    Article Google Scholar
  4. Perez CA, Salinas CA, Estevez PA, Valenzuela PM (2003) Genetic design of biologically inspired receptive fields for neural pattern recognition. IEEE Trans Syst Man Cybernetics Part B 33:258–270
    Article Google Scholar
  5. Funabiki N, Kitamichi J, Nishikawa S (1998) An evolutionary neural network approach for module orientation problems. IEEE Trans Syst Man Cybernetics Part B 28:849–855
    Article Google Scholar
  6. Baluja S (1996) Evolution of an artificial neural network based autonomous land vehicle controller. IEEE Trans Syst Man Cybernetics Part B 26:450–463
    Article Google Scholar
  7. Richards M, Moriarty D, Miikkulainen R (1997) Evolving neural networks to play Go. Appl Intell 8:85–96
    Article Google Scholar
  8. Agogino A, Stanley K, Mikkulainen R (2000), Online interactive neuro-evolution. Neural Process Lett 11:29–37
    Article Google Scholar
  9. Castillo-Valdivieso PA, Merelo JJ, Prieto A, Rojas I, Romero G (2002) Statistical analysis of the parameters of a neuro-genetic algorithm. IEEE Trans Neural Netw 13:1374–1394
    Article Google Scholar
  10. Stanley K, Miikkulainen R (2002) Evolving neural networks through augmenting topologies. Evol Comput 10:99–127
    Article Google Scholar
  11. Abbass HA (2003) Speeding up backpropagation using multiobjective evolutionary algorithms. Neural Comput 15:2705–2726
    Article MATH Google Scholar
  12. Bullinaria JA (2005) Evolving neural networks: is it really worth the effort? In: Proceedings of the European symposium on artificial neural networks, d-side, Evere Belgium, pp 267–272
  13. Bullinaria JA (2003) Evolving efficient learning algorithms for binary mappings. Neural Netw 16:793–800
    Article Google Scholar
  14. Bottou L, Le Cun Y (2004) Large scale online learning. In: Advances in neural information processing systems 16. MIT Press, Cambridge
    Google Scholar
  15. Bullinaria JA (2001) Simulating the evolution of modular neural systems. In: Proceedings of the twenty-third annual conference of the cognitive science society. Lawrence Erlbaum Associates, Mahwah, pp 146–151
  16. Bullinaria JA (2003) From biological models to the evolution of robot control systems. Phil Trans R Soc Lond A 361:2145–2164
    Article Google Scholar
  17. Seipone T, Bullinaria JA (2005) Evolving improved incremental learning schemes for neural network systems. In: Proceedings of the 2005 IEEE congress on evolutionary computing, IEEE, Piscataway, pp 273–280
  18. Eiben AE, Smith JE (2003) Introduction to evolutionary computing. Springer, Berlin
    MATH Google Scholar
  19. Whitley D (1989) The GENITOR algorithm and selection pressure: why rank-based allocation of reproductive trials is best. In: Schaffer JD (ed) Proceedings of the third international conference on genetic algorithms. Morgan Kaufmann, San Mateo, pp 116–123
  20. Syswerda G (1991) A study of reproduction in generational and steady-state genetic algorithms. In: Rawlins G (ed) Foundations of genetic algorithms. Morgan Kaufmann, San Mateo, pp 94–101
    Google Scholar
  21. De Jong KA, Sarma J (1993) Generation gaps revisited. In: Whitley LD (ed) Foundations of genetic algorithms 2. Morgan Kaufmann, San Mateo, pp 19–28
    Google Scholar
  22. Bullinaria JA (2004) Generational versus steady-state evolution for optimizing neural network learning. In: Proceedings of the international joint conference on neural networks, IEEE, Piscataway, pp 2297–2302
  23. Gould SJ, Eldredge N (1977) Punctuated equilibria: the tempo and mode of evolution reconsidered. Paleobiology 3:115–151
    Google Scholar
  24. Ursem RK (2002) Diversity-guided evolutionary algorithms. In: Merelo Guervós JJ et al (ed) Parallel problem solving from nature. Springer, Heidelberg, pp 462-471
    Google Scholar
  25. Lee Y, Oh SH, Kim MW (1992) An analysis of premature saturation in back propagation learning. Neural Netw 6:719–728
    Article Google Scholar
  26. Julesz B, Kovacs I (1995) Maturational windows and adult cortical plasticity. Addison-Wesley, Reading
    Google Scholar
  27. Bailey DB, Bruer JT, Symons F, Lichtman JW (2000) Critical thinking about critical periods. Brookes, Baltimore
    Google Scholar
  28. Jacobs RA (1988) Increased rates of convergence through learning rate adaptation. Neural Netw 1:295–307
    Article Google Scholar
  29. Hansen LK, Salamon P (1990) Neural network ensembles. IEEE Trans Pattern Anal Mach Intell 12:993–1000
    Article Google Scholar
  30. Battiti R, Colla AM (1994) Democracy in neural networks: voting schemes for classification. Neural Netw 7:691–709
    Article Google Scholar
  31. Yao X, Liu Y (1998) Making use of population information in evolutionary artificial neural networks. IEEE Trans Syst Man Cybernetics Part B 28:417–425
    Article Google Scholar
  32. Turner K, Ghosh J (1996) Error correlation and error reduction in ensemble classifiers. Conn Sci 8:385–403
    Article Google Scholar
  33. Bullinaria JA (2005) Evolved age dependent plasticity improves neural network performance. In: Proceedings of the fifth international conference on hybrid intelligent systems, IEEE, Piscataway, pp 79–84
  34. Bullinaria JA (2006) Ensemble techniques for avoiding poor performance in evolved neural networks. In: Proceedings of the international joint conference on neural networks, IEEE, Piscataway, pp 9857–9864

Download references

Acknowledgments

This paper brings together into a unified framework, and expands upon, a number of ideas and results previously scattered across several conference papers [12, 15, 17, 22, 33, 34].

Author information

Authors and Affiliations

  1. School of Computer Science, University of Birmingham, Birmingham, B15 2TT, UK
    John A. Bullinaria

Authors

  1. John A. Bullinaria

Corresponding author

Correspondence toJohn A. Bullinaria.

Rights and permissions

About this article

Cite this article

Bullinaria, J.A. Using evolution to improve neural network learning: pitfalls and solutions.Neural Comput & Applic 16, 209–226 (2007). https://doi.org/10.1007/s00521-007-0087-9

Download citation

Keywords