Abstract
Training multilayer feedforward neural networks corresponds to the global minimization of the network error function. To address this problem we utilize the Snyman and Fatti [1] approach by considering a system of second order differential equations of the form, ẍ\(=-\nabla E(x)\), where x is the vector of network weights and \(\nabla E\) is the gradient of the network error function E. Equilibrium points of the above system of differential equations correspond to optimizers of the network error function. The proposed approach is described and experimental results are discussed.
Keywords
This work is partially supported by the “Pythagoras” research grant awarded by the Greek Ministry of Education and Religious Affairs and the European Union.
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
Snyman, J., Fatti, L.: A multi–start global minimization algorithm with dynamic search trajectories. JOTA 54, 121–141 (1987)
Törn, A., Žilinskas, A.: Global optimization. In: Törn, A., Žilinskas, A. (eds.) Global Optimization. LNCS, vol. 350, Springer, Heidelberg (1989)
Incerti, S., Parisi, V., Zirilli, F.: A new method for solving nonlinear simultaneous equations. SIAM J. Num. Anal 16, 779–789 (1979)
Inomata, S., Cumada, M.: On the golf method. Bulletin of the Electronical Laboratory 25, 495–512 (1964)
Zhidkov, N., Shchdrin, B.: On the search of minimum of a function of several variables. Computing methods and Programming 10, 203–210 (1978)
Pshenichnyi, B., Marchenko, D.: On one approach to the search for the global minimum. Optimal Decision theory 2, 3–12 (1967)
Griewank, A.: Generalized descnet for global optimization. JOTA 34, 11–39 (1981)
Petalas, Y.G., Tasoulis, D.K., Vrahatis, M.N.: Trajectory methods for neural network training. In: Proceedings of the IASTED International Conference on Artificial Intelligence and Applications (AIA 2004), vol. 1, pp. 400–408. ACTA press (2004)
Armijo, L.: Minimization of function having lipschitz continuous first partial derivatives. Pac J. Math. 16, 1–3 (1966)
Magoulas, G., Vrahatis, M.N., Androulakis, G.: Effective backpropagation training with variable stepsize. Neural Networks 10, 69–82 (1997)
Magoulas, G., Vrahatis, M.N., Androulakis, G.: Increasing the convergence rate of the error backpropagation algorithm by learning rate adaptation methods. Neural Computation 11, 1769–1796 (1999)
Vogl, T., Mangis, J., Rigler, A., Zink, W., Alkon, D.: Accelerating the convergence of the back-propagation method. Biol. Cybern. 59, 257–263 (1988)
Rao, S.: Optimization theory and Applications. Wiley Eastern Limited, Chichester (1992)
Riedmiller, M., Braun, H.: A direct adaptive method for faster backpropagation learning: The rprop algorithm. In: Proceedings of the IEEE International Conference on Neural Networks, San Francisco, CA, pp. 586–591 (1993)
Igel, C., Hüsken, M.: Improving the Rprop learning algorithm. In: Bothe, H., Rojas, R. (eds.) Proceedings of the Second International ICSC Symposium on Neural Computation (NC 2000), pp. 115–121. ICSC Academic Press, London (2000)
Prechelt, L.: Proben1: A set of neural network benchmark problems and benchmarking rules. Technical Report 21/94 (1994)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2004 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Petalas, Y.G., Tasoulis, D.K., Vrahatis, M.N. (2004). Dynamic Search Trajectory Methods for Neural Network Training. In: Rutkowski, L., Siekmann, J.H., Tadeusiewicz, R., Zadeh, L.A. (eds) Artificial Intelligence and Soft Computing - ICAISC 2004. ICAISC 2004. Lecture Notes in Computer Science(), vol 3070. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-24844-6_32
Download citation
DOI: https://doi.org/10.1007/978-3-540-24844-6_32
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-22123-4
Online ISBN: 978-3-540-24844-6
eBook Packages: Springer Book Archive