Effects of global perturbations on learning capability in a CMOS analogue implementation of synchronous Boltzmann machine

  • Kurosh Madani
  • Ghislain de Tremiolles
Artificial Neural Nets Simulation and Implementation
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1607)


All of the presented implementations of Artificial Neural Networks (A.N.N.) have been supposed to be working in ideal conditions, however, real applications will be subject to local and global perturbations. Since 1994, we have investigated the behaviour modelling of electronic A.N.N. with global perturbation conditions. We have scrutinised the behaviour analysis of a CMOS analogue implementation of synchronous Boltzmann Machine model with both ambient temperature and electrical perturbation. In this paper we present, using our model, the analysis of these global perturbations effects on learning capability of the above mentioned CMOS based analogue implementation. Simulation and experimental results have been exposed validating our concepts.

Key Words

Global Perturbations Neural network Learning capability Modelling Experimental validation 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    C.A. Mead, “Analogue VLSI and neural systems”, Addison Wesley 1989.Google Scholar
  2. [2]
    R.F. Lyon and C.A. Mead, “An Analog electronic cochlea”, IEEE transactions on Acoustic, Speech and signal Processing, Vol. 36, PP 1119–1134, 1988.MATHCrossRefGoogle Scholar
  3. [3]
    L. Jackel, “Electronic neural networks”. In NATO ARW, Neuro-algorithms, architecture and applications, Les Arcs, 1989.Google Scholar
  4. [4]
    M. MAYOUBI, M. SCHAFER, S. SINSEL, “Dynamic Neural Units for Non-linear Dynamic Systems Identification”, From Natural to Artificial Neural Computation, LNCS Vol. 930, Springer Verlag, pp. 1045–1051,, 1995.Google Scholar
  5. [5]
    M. CHIABERGE, L. M. REYNERI, “Cintia: A Neuro-Fuzzy Real-Time Controller for Law-Power Embedded Systems”, IEEE Micro Vol. 15, pp. 40–47, June 1995.CrossRefGoogle Scholar
  6. [6]
    G. MERCIER, K. MADANI, “CMAC Real-Time Adaptive Control Implementation on a DSP Based Card”,, From Natural to Artificial Neural Computation, LNCS, Vol. 930, Springer Verlag, pp. 1114–1120, 1995.Google Scholar
  7. [7]
    G. Bugmann, P. Sojka, M. Reiss, M. Plumbly, J. Taylor, “Direct Approaches to Improving the robustness of Multilayer Neural Networks”, Artificial Neural Networks 2, Elsevier science Pub, 1992.Google Scholar
  8. [8]
    J.J. Hopfield, “Neurons with graded response have collective computational properties like those of two state neurones”, Proceedings of the national Academy of science of U.S.A., vol 81 pp 3088–3092, 1984.CrossRefGoogle Scholar
  9. [9]
    J.L. WYATT and D.L. STANDLEY, “Circuit design criteria for stable lateral inhibition neural networks “In IEEE International Symposium Circuits and systems, IEEE pp 997–1000, June 1988.Google Scholar
  10. [10]
    M.A. Sivilotti, M.R. EMERLING and C.A. Mead, “VLSI Architectures for implementation of Neural Network”. In AIP conference Proceedings on Neural Network for computing, J.S. DENKER, American Institute of physic, Snowbird, UTAH pp408–413, 1986.Google Scholar
  11. [11]
    M. VERLEYSEN and P. JESPERS, “precision of sum-of-product in Analog Neural Network”. In Proceedings of the first International workshop on Microelectronics for Neural Networks, Dortmund, RFA, June 1990.Google Scholar
  12. [12]
    K. MADANI, I. BERECHET, “Temperature Perturbation Effects on Image Processing Dedicated stochastic Artificial Neural Networks”, SPIE SYMPOSIUM ON ELECTRONIC IMAGING: Science and Technology, San Jose California-U.S.A., February 6–10 1994.Google Scholar
  13. [13]
    G.E. Hinton and T.J. Sejnowski, “learning in Boltzmann machines”. In Cognitive 85, PARIS, PP 283–290, 1985.Google Scholar
  14. [14]
    R. Azencott, “Synchronous Boltzmann Machines and their learning algorithms”. In NATO ARW, Springer-Verlag, les arcs, February 1989.Google Scholar
  15. [15]
    P. GARDA and E. BELHAIRE, “An Analog chip set with digital I/O for synchronous Boltzmann Machine.” “In VLSI for Artificial Intelligence and Neural Network Kluwer Academic, J.G. Delgado-frias and W.R. Moore, BOSTON, 1990.Google Scholar
  16. [16]
    V. LAFARGUE, “Contribution à la réalisation électronique de Réseaux de Neurones formels: Intégration mixte de l'apprentissage des machines de Boltzmann”; Ph. D. Report, thèse de doctorat en science de l'université PARIS XI, Orsay, January 1993.Google Scholar
  17. [17]
    K. MADANI, I. BERECHET, G. DE TREMIOLLES, “Analysis of limitations in Analog Implementation of stochastic Artificial Neural Network V, ORLANDO, FLORIDA, U.S.A., 4–8 pril 1994.Google Scholar
  18. [18]
    E. BELHAIR, “Contribution à la réalisation électronique de réseaux de Neurones Formels Intégration Analogique d'une machine de BOLTZMANN”; ph.D. report, thèse de doctorat en science de l'université Paris XI, Orsay February 1992.Google Scholar
  19. [19]
    Y.P. TSIVIDIS, “Operation and Modelling of the MOS Transistor”, Mc Graw Hill, 1988, PP148.Google Scholar
  20. [20]
    S.M. SZE “physics of Semiconductor Devices”. Wiley, 1981.Google Scholar
  21. [21]
    K. MADANI, G. DE TREMIOLLES, Perturbation Effects Analysis in Analogue Implementation of a stochastic Artificial Neural Network, SPIE International AeroSense'96 Symposium—Applications and Science of Artificial Neural Networks, Orlando, Florida, USA, 08–12 May 1996.Google Scholar
  22. [22]
    K. MADANI, G. DE TREMIOLLES, Global Perturbation Effects Analysis in a CMOS Analogue Implementation of Synchronous Boltzmann Machine, 3-rd International Workshop on Thermal Investigations of Integrated Circuits and Microstructures, IEEE-CNRS, Cannes—Côte d'Azur, September 21–23, 1997.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1999

Authors and Affiliations

  • Kurosh Madani
    • 1
  • Ghislain de Tremiolles
    • 1
  1. 1.Division Réseaux NeuronauxLeriss-Université Paris XII, I.U.T. De SenartLieusaintFrance

Personalised recommendations