Abstract
An algorithm of learning in multilayer threshold nets without feedbacks is proposed. The net is. built of threshold elements with binary inputs. During a learning process each input vector x is accompanied by a teacher's decision ω (ωε{1,...,M}). The pairs (x[n], ω[n]) appear in successive steps independently according to some unknown stationary distribution p(x,ω). The problem of learning of a threshold net has been decomposed to a series of problems of learning of the threshold elements. The proposed learning algorithm of the threshold elements has a perceptron-like form. It was proven that a decision rule of the threshold net stabilizes after a finite number of steps. For definite classes {p(x, ω)} K* of distributions p(x,ω), an optimal decision rule stabilizes after a finite number of steps. These classes {p(x, ω)} K* also contain distributions describing learning processes with perturbations.
Similar content being viewed by others
References
Alder, M.D.: A convergence theorem for hierarchies of model neurones. SIAM J. Comput 4, 192–201 (1975)
Amari, S.: Neural theory of association and concept-formation. Biol. Cybernetics 26, 175–185 (1977)
Bobrowski, L.: Learning of a threshold element in the case of binary stochastic inputs (in Polish). Prace IOK37, 1–21 (1976)
Bobrowski, L.: Asymptotical characteristics of a certain learning algorithm in the stochastic case (in Polish). Arch. Autom. i Telemech. XXII 3, 239–250 (1977a)
Bobrowski, L.: The selected problems of pattern recognition learning in threshold nets (in Polish). Ph. D. Thesis, IBS, PAN, Warsaw (1977b)
Caianiello, E.R.: Methods of analysis of neural networks. Biol. Cybernetics 22, 1–6 (1976)
Gawroński, R.: The choice of the structure of neuronlike nets for the applications to preliminary selection of information. The Scientific Papers of the Institute of the Technical Cybernetics (Wroclaw) 16, 53–72 (1974)
Newelson, M.B.: Chasminskij, P.Z.: Stochastic approximation and recursive estimation (in Russian). Moscow: Nauka 1972
Nilsson, N.J.: Learning machines. New York: McGraw Hill 1965
Rosenblatt, F.: Principles of neurodynamics. Washington: Spartan Books 1962
Wapnik, W.N.: Czerwonenkis, A.J.: Theory of pattern recognition. (in Russian). Moscow: Nauka 1974
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Bobrowski, L. Learning processes in multilayer threshold nets. Biol. Cybernetics 31, 1–6 (1978). https://doi.org/10.1007/BF00337365
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF00337365