Abstract
We examine the storage capacity for the binary perceptron using simulated annealing. In particular, we clarify the connection between the computational complexity of learning algorithms and the attained storage capacity. From finite-size studies we obtain a critical storage capacity,α c (κ)=0.8331±0.0016, in good agreement with the replica analysis of Krauth and Mézard. However, we demonstrate that a polynomial time cooling schedule yields a vanishing storage capacity in the thermodynamic limit as predicted by the dynamical theory of Horner. Nonetheless, we show these two results may be reconciled by explicitly verifying that the learning problem for the binary perceptron is NP-complete. This investigation has been made possible by the development of an accelerated annealing algorithm.
Similar content being viewed by others
References
Minsky, M., Pappert, S.: Perceptrons: An introduction to computational geometry. Cambridge: M.I.T. Press 1969
Hertz, J., Krogh, A., Palmer, R.G.: Introduction to the theory of neutral computation. Reading, Mass: Addison-Wesley 1991
Krauth, W., Mézard, M.: J. Phys. A20, 1745 (1987)
Anlauf, J.K., Biehl, M.: Europhys. Lett.10, 687 (1989)
Gutfreud, H., Stein, Y.: J. Phys. A233, 2613 (1990)
Press, W.H., Flannery, B.P., Teukolsky, S.A., Vetterling, W.T.: Numerical recipes: the art of scientific computing, C.U.P. 1986
Karmarker, N.: Combinatorica4, 373 (1984)
Pitt, L., Vailant, L.G.: J. ACM35(4), 965 (1988)
Cover, T.M.: IEEE Trans. Comput. EC-14, 326 (1965)
Gardner, E.: J. Phys. A21, 257 (1988)
Mézard, M., Parisi, G., Virasoro, M.A.: Spin glass theory and beyond. Singapore: World Scientific 1987
Gardner, E., Derrida, B.: J. Phys. A21, 271 (1988)
Krauth, W., Mézard, M.: J. Phys. A22, 3057 (1989)
Derrida, B.: Phys. Rev. B24, 2631 (1981)
Gross, D.J., Mézard, M.: Nucl. Phys. B240, 431 (1984)
Krauth, W., Opper, M.: J. Phys. A22 L519 (1989)
Horner, H.: Z. Phys. B86, 291 (1992)
Sherrington, D., Kirkpatrick, S.: Phys. Rev. Lett.35, 1792 (1975)
Fu, Y., Lectures in the sciences of complexity. Stein, D. (ed.), p. 815, Reading, Mass: Addison-Wesley 1989
Köhler, H., Diedreich, S., Kinzel, W., Opper, M.: Z. Phys. B78, 333 (1990)
Amaldi, E., Nicolis, S.: J. Phys.50, 2333 (1989)
Perez, C.J., Carrabina, J., Valderrama, E.: Network3, 165 (1992)
Fontanari, J.F., Köberle, R.: J. Phys.51, 1403 (1990)
Köhler, H.: J. Phys. A23, L1265 (1990)
Kirkpatrick, S., Gelatt, Jr., C.D., Vecchi, M.P.: Science220, 671 (1983)
Harland, J.R., Salamon, P.: Nucl. Phys. B (Proc. Suppl)5A, 109 (1988)
Metropolis, N., Rosenbluth, A.W., Rosenbluth, M.N., Teller, A.H., Teller: J. Chem. Phys.21, 1087 (1953)
Nulton, J.D., Salamon, P.: Phys. Rev. A37, 1351 (1988)
Horner, H.: (Private communication)
Derrida, B., Griffiths, R.B., Prügel-Bennet, A.: J. Phys. A24, 4907 (1991)
Gardner, E., Derrida, B.: J. Phys. A22, 1983 (1989)
Györghi, G.: J. Phys. Rev. A41, 7097 (1990)
Sompolinsky, H., Tishby, N., Seung, H.S.: Phys. Rev. Lett.65, 1683 (1991)
Horner, H.: Z. Phys. B87, 371 (1992)
Kühn, R.: (Private communication)
Tausworth, R.C.: Math. Commut.19, 201 (1965)
Ahrens, J.H., Dieter, U., Grube, A.: Computing6, 121 (1970)
Dueck, G., Scheuer, T.: J. Comput. Phys.90, 161 (1990)
Creutz, M.: Phys. Rev. Lett.50, 1411 (1983)
Guo, H., Zuckermann, M., Harris, R., Grant, M.: Phys. Scri. T38, 40 (1991)
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Patel, H.K. Computational complexity, learning rules and storage capacities: A Monte Carlo study for the binary perceptron. Z. Physik B - Condensed Matter 91, 257–266 (1993). https://doi.org/10.1007/BF01315244
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF01315244