Abstract
Hadamard theory is shown to play an important role in the generation of Boolean decision functions, a fundamental tool in the field of artificial neural network design. Based on a group-theoretic introduction of a complete set of Hadamard vectors, whose matrices are of the order of a power of two, we classify subsets according to the degree of their linear dependence. We show in the thermodynamic limit that essentially the whole Hadamard space is occupied by representatives with defect not exceeding two or three.
References
A. Hedayat and W. D. Wallis, Hadamard matrices and their applications,Ann. Stat. 6:1184–1237 (1978).
W. K. Pratt, J. Kane, and H. C. Andrees, Hadamard transform image coding,Proc. IEEE 57(1):58–69 (1969).
E. C. Posner, Combinatorial structure in planetary reconnaissance, inProceedings Symposium on Error Correcting Codes (Mathematics Research Center, University of Wisconsin Press, Madison, Wisconsin, 1968).
M. H. Tai, M. Harit, and N. J. A. Sloane, Errors in Hadamard spectroscopy or imaging caused by imperfect masks,Appl. Opt. 14(11):2678–2686 (1975).
D. E. Goldberg, Genetic algorithms and Walsh functions: Part I, A gentle introduction,Complex Syst. 129:129–152 (1989).
O. Krisement, A Hopfield model with Hadamard prototypes,Z. Phys. B 80:415–421 (1990).
R. Folk and A. Kartashov, Equivalence classes of Hadamard pattern sets for Hebbian networks,Neural Network World, submitted.
J. V. Brawley, Jr., and Petr Lisonek, Counting equivalence classes of Hadamard pattern sets by group action methods, RISC-Linz Report Series No. 92-28 (1992).
K. E. Kürten, Information specific design in high-order networks, inComplexity in Physics and Technology, M. S. Garrido and R. Vilela Mendes, eds. (World Scientific, Singapore, 1992), pp. 167–180.
S. Kauffman,J. Theor. Biol. 22:437 (1969).
E. R. Caianiello, Neuronic equations and their solutions, inEvolution, Learning and Cognition, Y. C. Lee, ed. (World Scientific, Singapore, 1988).
R. D. Carmichael,Introduction to the Theory of Groups of Finite Order (Dover, New York).
J. Cigler, Uber die Anzahl erzeugender Mengen in endlichen Vektorräumen,Anz. Öst. Akad. Wiss., in press.
K. E. Kürten, Adaptive architectures for Hebbian network models.J. Phys. I (Paris)2:615–624 (1992).
K. E. Kürten, Optimal dilution of a Hopfield network, inNeurodynamics, F. Pasemann and H. D. Doebner, eds. (World Scientific, Singapore, 1991), pp. 211–221.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Kürten, K.E., Klingen, N. Hadamard design and artificial neural nets. J Stat Phys 71, 327–339 (1993). https://doi.org/10.1007/BF01048103
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF01048103