Summary
The learning process of multilayer neural networks can be considered as a multistage optimal control process. We introduce a gain parameter into the models of neurons. Setting the parameter to a small value makes the neuron model “almost linear”, and the learning process problem can be solved using computational tools specified for linear-quadratic systems optimization. In Part 1 the continuation methodology is applied for changing the gain parameter in order to reach 1.0. In Part 2, by considering the gain parameter as an additional control variable, the optimal value of the parameter can be found. The methodology we propose to call the heuristic dynamic programming.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsPreview
Unable to display preview. Download preview PDF.
References
Aarts, E., Laarhoven, P. (1987). Simulated Annealing: Theory and Practice. Wiley and sons, N-Y.
Avila, J. H., (1974). The Feasibility of Continuation Methods for Nonlinear Equations. SIAM J. Numer. Anal., vol. 11, No. 1.
Davidenko, D. (1953). On the Approximate Solution of a System of Nonlinear Equations. Dokl. Akad. SSSR, 88, 601–604. (in Russian).
Krawczak, M. and Mizukami, K. (1994). The control theory approach to perceptron learning process. IEE of Japan, Okayama.
Krawczak, M. (2000a). Backpropagation versus dynamic programming approach. Bulletin of Polish Academy of Sciences. Vol. 48, No. 2, pp. 167–180.
Krawczak, M. (2000b). Feedforward neural networks learning by continuation method. Proc. IIZUKA2000. Fukuoka — Japan, October 2000.
Krawczak, M. (2001a). Parameterization of neural networks learning. Proc. 14 th Int. Conference on Systems Science, Wroclaw, Poland, September 2001.
Krawczak, M. (2001b). Neural networks learning as a particular optimal control problem. Proc. 7 th Int. Conference on Methods and Models in Automation and Robotics, IEEE, Mi@dzyzdroje, Poland, August 2001.
Ortega J. M., Rheinboldt, W. C. (1970). Iterative Solution of Nonlinear Equations in Several Variables. Academic Pres, N-Y, S. Francisco, London.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2003 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Krawczak, M. (2003). Heuristic Dynamic Programming for Neural Networks Learning Part 1: Learning as a Control Problem. In: Rutkowski, L., Kacprzyk, J. (eds) Neural Networks and Soft Computing. Advances in Soft Computing, vol 19. Physica, Heidelberg. https://doi.org/10.1007/978-3-7908-1902-1_30
Download citation
DOI: https://doi.org/10.1007/978-3-7908-1902-1_30
Publisher Name: Physica, Heidelberg
Print ISBN: 978-3-7908-0005-0
Online ISBN: 978-3-7908-1902-1
eBook Packages: Springer Book Archive