Abstract
Within the context of a particular statistical discrimination task, we make a quantitative comparison between the performance of a feed-forward neural network and the information-theoretic optimal performance. We also address the ability of such networks to generalize and the effect of network architecture on performance.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
W. Bialek and A. Zee,Phys. Rev. Lett. 58:741 (1987).
W. Bialek and A. Zee,Phys. Rev. Lett. 61:1512 (1988).
D. Waltz and J. A. Feldman, eds.,Connectionist Models and Their Implications (Ablex, Norwood, New Jersey, 1988), and references therein.
D. E. Rumelhart, G. E. Hinton, and R. J. Williams, inParallel Distributed Processing: Explorations in the Microstructure of Cognition, D. E. Rumelhart and J. L. McClelland, eds. (Bradford, Cambridge, 1986).
T. J. Sejnowski and C. R. Rosenberg,Complex Systems 1:145–168 (1987).
R. T. Scalettar and A. Zee,Biol. Cybernet. 58:193 (1988).
D. M. Green and J. A. Siwets,Signal Detection Theory in Psychophysics (Krieger, New York, 1966).
N. Qian and T. Sejnowski,J. Mol. Biol. 202, in press.
E. Baum and F. Wilczek, inNeural Information Processing Systems, D. Z. Anderson, ed. (AIP Press, New York, 1988), pp. 52–61.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Bialek, W., Scalettar, R. & Zee, A. Optimal performance of a feed-forward network at statistical discrimination tasks. J Stat Phys 57, 141–156 (1989). https://doi.org/10.1007/BF01023637
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF01023637