Abstract
Learning from examples by perceptron withN binary synaptic couplings is investigated within dynamic mean field theory. This applies to learning by simulated annealing, which is a polynomial algorithm for finite cooling rates. For examples created by a teacher perceptron of the same type, a discontinuous freezing transition occurs for a training set of size αN with α<1.58 at a temperature where the entropy is still positive. The resulting perceptrons have finite training and generalization error. For α>1.58 the couplings of the teacher are found by the above process. This work extends previous investigations on a binary perceptron trained with random patterns.
Similar content being viewed by others
References
M. Minsky, S. Papert: Perceptrons. An introduction to computational geometry. Cambridge MA: MIT Press 1969
F. Rosenblatt: Principles of neurodynamics. New York: Spartan Books 1962
D. Amit: Modelling brain functions. Cambridge: Cambridge University Press 1989
J. Hertz, A. Krogh, R.G. Palmer: Introduction to the theory of neural computing. Reading, Mass: Addison Wesley 1991
E. Gardner: J. Phys. A21, 257 (1988)
E. Gardner, B. Derrida: J. Phys. A22, 1983 (1989)
G. Györgyi: Phys. Rev. A41, 7097 (1990)
H. Sompolinsky, N. Tishby, H.S. Seung: Phys. Rev. Lett.65, 1683 (1990); H.S. Seung, H. Sompolinsky, N. Tishby: Preprint (1991)
B. Derrida, R.B. Griffiths, A. Prügel-Bennett: J. Phys. A24, 4907 (1991)
E.B. Baum, Y.D. Lyuu: Neural Comput.3, 386 (1991)
S. Diederich, M. Opper: Phys. Rev. Lett.58, 949 (1987)
W. Kraut, M. Mezard: J. Phys. A20, L745 (1987)
J.K. Anlauf, M. Biehl: Europhys Lett.10, 687 (1989)
E. Gardner, B. Derrida: J. Phys. A21, 271 (1988),23, 1983 (1989)
W. Kraut, M. Mezard: J. Phys. (Paris)50, 3054 (1989)
W. Kraut, M. Opper: J. Phys. A22, L519 (1989)
H. Gutfreund, Y. Stein: J. Phys. A23, 2613 (1990)
H. Horner: Z. Phys. B — Condensed Matter86, 291 (1992)
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Horner, H. Dynamics of learning and generalization in a binary perceptron model. Z. Physik B - Condensed Matter 87, 371–376 (1992). https://doi.org/10.1007/BF01309290
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF01309290