Advertisement

Neural Networks Learning Rules for Control: Uniform Dynamic Backpropagation, Heavy Adaptive Learning Rule

  • Nicolas Seube
Part of the The Springer International Series in Engineering and Computer Science book series (SECS, volume 202)

Abstract

This paper presents two original learning rules, based on a certain knowledge on a plant dynamics, aiming to find regulation laws for controlled systems. The Uniform Dynamic BackPropagation rule is based on a non regular optimization scheme (subgradient algorithm), and is devoted to the minimisation of a Min Max criterion, in a neural network synaptic matrix space. The Heavy Adaptive learning rule is a continuous hebbian learning rule, enabling a network to adaptively learn a feedback control map. Simulation results of the two rules are presented for the control of an underwater vehicle.

Keywords

Learning Rule Underwater Vehicle Subgradient Algorithm Surging Velocity Clarke Derivative 
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]
    J-P Aubin. Mathematical Methods of Artificial Intelligence. Université de Paris Dauphine, 1990. Lecture Notes.Google Scholar
  2. [2]
    J-P Aubin. Viability theory. Birkhauser, 1991.Google Scholar
  3. [3]
    J-P Aubin and H. Frankowska. Set valued analysis. Birkhauser, 1990.Google Scholar
  4. [4]
    A.G. Barto, R.S. Sutton, and C.W. Anderson. Neuronlike elements that can solve difficult learning control problems. IEEE Trans. on Systems, Man, and Cybernetics,1983.Google Scholar
  5. [5]
    G. CYBENKO. Approximation by superposition of a sigmoidal function. Math. Control Signal Systems, 2:315–341, 1989.MathSciNetCrossRefGoogle Scholar
  6. [6]
    A. Guez and J. Selinsky. A trainable neuromorphic controller. Journal of Robotic Systems, 4(5), 1988.CrossRefGoogle Scholar
  7. [7]
    M.I Jordan, Forward models: learning with a distal teacher. June 1990. Preprint.Google Scholar
  8. [8]
    M.I Jordan and R.A. Jacobs. Learning to Control an Unstable System. With Forward Moddeling. 1990.Google Scholar
  9. [9]
    M. Saerens and A. Soquet. A neural controller based on back-propagation algorithm. ln First IEE International Conference On Artificial Neural Networks, pages 211–215, London, 1989.Google Scholar
  10. [10]
    N. Seube. Apprentissage de lois de contrôle régulant des contraintes sur l’état par réseaux de neurones. In Comptes Rendus de l’Académie des Sciences de Paris, Série I, pages 445–450, 1991.Google Scholar
  11. [11]
    N. Seube. Construction of learning rule in neural networks that can find viable regulation laws to control problems by self-organization. In International Neural Networks Conference, pages 209–212, Paris, 1990.Google Scholar
  12. [12]
    N. Seube. Neural network learning rule for control: application to auv tracking control. In Neural Network For Ocean Engineering Workshop, pages 185–196, Washington DC, August 1991.Google Scholar
  13. [13]
    N. Seube and J-C Macias. Design of neural network learning rules for viable feedback laws. In European Control Conference, pages 1241–1246, Grenoble, July 1991.Google Scholar
  14. [14]
    P.J Verbos. Back-propagation through time: what it does and how to do it. Proceeding of the IEEE, 78(10), October 1990.Google Scholar
  15. [15]
    B. Widrow and D. Nguyen. The truck backer-upper: an example of self-learning in neural networks. In Joint International Conference on Neural Networks, 1989.Google Scholar
  16. [16]
    J. Yuh. A neural net controller for underwater robotic vehicules. IEEE Journal of Oceanic Engeneering, 15(3):161–166, 1990.CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media New York 1993

Authors and Affiliations

  • Nicolas Seube
    • 1
  1. 1.Thomson Sintra Activités Sous-Marines 1Arcueil CedexFrance

Personalised recommendations