Advertisement

Discretization of ISO-Learning and ICO-Learning to Be Included into Reactive Neural Networks for a Robotics Simulator

  • José M. Cuadra Troncoso
  • José R. Álvarez Sánchez
  • Félix de la Paz López
Part of the Lecture Notes in Computer Science book series (LNCS, volume 4528)

Abstract

Isotropic Sequence Order learning (ISO-learning) and Input Correlation Only learning (ICO-learning) are unsupervised neural algorithms to learn temporal differences. The use of devices implementing this algorithms by simulation in reactive neural networks is proposed. We have applied several modifications to original rules: weights sign restriction, to adequate ISO-learning and ICO-learning devices outputs to the usually predefined kinds of connections (excitatory/inhibitory) used in neural networks, and decay term inclusion for weights stabilization. Original experiments with these algorithms are replicated as accurate as possible with a simulated robot and a discretization of the algorithms. Results are similar to those obtained in original experiments with analogue devices.

Keywords

Unconditioned Stimulus Original Experiment Hebbian Learning Unconditioned Response Robot Behavior 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Oppenheim, A.V., Willsky, A.S., Nawab, S.H.: Signals and systems, 2nd edn. Prentice-Hall, Englewood Cliffs (Aug. 1996)Google Scholar
  2. 2.
    Braitenberg, V.: Vehicles: experiments in synthetic psychology. MIT Press, Cambridge (1986)Google Scholar
  3. 3.
    Gerstner, W., Kistler, W.M.: Mathematical formulations of Hebbian learning. Biological Cybernetics 87, 404–415 (2002)zbMATHCrossRefGoogle Scholar
  4. 4.
    Kempter, R., Gerstner, W., van Hammen, L.L.: Intrinsic stabilization of output rates by spike-based Hebbian learning. Neural Computation 13, 2709–2741 (2001)zbMATHCrossRefGoogle Scholar
  5. 5.
    Porr, B., von Ferber, C., Wörgötter, F.: ISO-learning aproximates a solution to the inverse-controller problem in an unsupervised behavioural paradigm. Neural Computation 15, 865–884 (2003)zbMATHCrossRefGoogle Scholar
  6. 6.
    Porr, B., Wörgötter, F.: Isotropic sequence order learning. Neural Computation 15, 831–864 (2003)zbMATHCrossRefGoogle Scholar
  7. 7.
    Porr, B., Wörgötter, F.: Isotropic sequence order learning in a closed-loop behavioural system. Roy. Soc. Phil. Trans. Mathematical, Physical & Engineering Sciences 361(1811), 2225–2244 (2003)CrossRefGoogle Scholar
  8. 8.
    Porr, B.: Sequence-Learning in a Self-Referential Closed-Loop Behavioural System. PhD thesis, Stirling University (May 2003)Google Scholar
  9. 9.
    Porr, B., Wörgötter, F.: Temporal Hebbian learning in rate-coded neural networks: A theoretical approach towards classical conditioning. In: Dorffner, G., Bischof, H., Hornik, K. (eds.) ICANN 2001. LNCS, vol. 2130, pp. 1115–1120. Springer, Heidelberg (2001)CrossRefGoogle Scholar
  10. 10.
    Porr, B., Wörgötter, F.: Strongly improved stability and faster convergence of temporal sequence learning by utilising input correlations only. Neural Computation 18(6), 1380–1412 (2006)zbMATHCrossRefGoogle Scholar

Copyright information

© Springer Berlin Heidelberg 2007

Authors and Affiliations

  • José M. Cuadra Troncoso
    • 1
  • José R. Álvarez Sánchez
    • 1
  • Félix de la Paz López
    • 1
  1. 1.Departamento de Inteligencia Artificial, UNEDSpain

Personalised recommendations