Advertisement

On the computational efficiency of symmetric neural networks

  • Juraj Wiedermann
Communications
Part of the Lecture Notes in Computer Science book series (LNCS, volume 379)

Abstract

An open problem concerning the computational power of neural networks with symmetric weights is solved. It is shown that these networks possess the same computational power as general networks with asymmetric weights — i.e. these networks can compute any recursive function. The computations of these networks can be described as a minimization process of a certain energy function; it is shown that for uninitialized symmetric neural networks this process presents a PSPACE-complete problem.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. [1]
    Ackley, D. N.-Hinton, G. E.-Sejnowski, T. I.: A Learning Algorithm for Boltzmann Machines. Cognitive Science 9, 1985, pp. 147–169Google Scholar
  2. [2]
    Barahona, F.: On the Computational Complexity of Ising Spin Glass Models. J. Phys. A. 15, 1982, pp. 3241–3253Google Scholar
  3. [3]
    Chandra, A. K.-Stockmeyer, L. I.-Vishkin, U.: Constant Depth Reducibility. SIAM J. Comput. Vol. 15, No. 3, 1984, pp. 423–432Google Scholar
  4. [4]
    Egecioglu, O.-Smith, T. R.-Moody, I.: Computable Functions and Complexity in Neural Networks. Tech. Rep. ITP-124, University of California, Santa Barbara, 1986Google Scholar
  5. [5]
    Feldman, J. A.: Energy and the Behavior of Connectionist Models. Tech. Rep. TR-155, University of Rochester, Nov. 1985Google Scholar
  6. [6]
    Hopfield, J. J.: Neural Networks and Physical Systems with Emergent Collective Computational Abilities. Proc. Natl. Acad. Sci. U.S.A. 79, 1982, pp. 2554–2558Google Scholar
  7. [7]
    Hopfield, J. J.: Neurons with Graded Response Have Collective Computational Properties Like Those of Two-state Neurons. Proc. Natl. Acad. Sci. U.S.A., 1984, pp. 3088–3092Google Scholar
  8. [8]
    Hopfield, J. J.-Tank, D. W.: ”Neural” Computations of Decisions in Optimization Problems. Biol. Cybern. 52, 1985, pp. 141–152Google Scholar
  9. [9]
    Minsky, M.: Computation. Finite and Infinite Machines. Prentice Hall, Englewood Cliffs, NJ, 1967Google Scholar
  10. [10]
    Parberry, I.-Schnitger, G.: Parallel Computation with Threshold Functions. JCSS 36, 1988, pp. 278–302Google Scholar
  11. [11]
    Robson, J. M.: Linear Size Formulas for Non-deterministic Single Tape Computations. Proc. 11-th Australian Comp. Sci. Conference, Brisbane, Feb. 3–5, 1988Google Scholar
  12. [12]
    van Emde Boas, P.: Machine Models and Simulations. ITLI Prepublication Series of Computation and Complexity Theory CT-88-95, University of Amsterdam, 1988Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1989

Authors and Affiliations

  • Juraj Wiedermann
    • 1
  1. 1.VUSEI-ARBratislavaCzechoslovakia

Personalised recommendations