Abstract
The selection of an adequate hidden structure of a feedforward neural network is a very important issue of its design. When the hidden structure of the network is too large and complex for the model being developed, the network may tend to memorize input and output sets rather than learning relationships between them. In addition, training time will significantly increase when the network is unnecessarily large. We propose two methods to optimize the size of feedforward neural networks using orthogonal transformations. These two approaches avoid the retraining process of the reduced-size network, which is necessary in any pruning technique.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
References
P. Bachiller, R.M. Pérez, P. Martínez and P.L. Aguilar: A method based on orthogonal transformation for the design of optimal feedforward network architecture. Proceedings of the 3rd International Meeting on Vector and Parallel Processing (VECPAR’98), 1998, pp. 541–552.
G. Castellano, A. M. Fanelli and M. Pelillo: An iterative pruning algorithm for feedforward neural networks. IEEE Transactions on Neural Networks, vol. 8, no. 3,1997, pp. 519–531.
G. H. Golub and C. F. Van Loan: Matrix computations. Baltimore, MD: John Hopkins Univ. Press, 1989.
P. Kanjilal and N. Banerjee: On the application of orthogonal transformation for the design and analysis of feedforward networks. IEEE Transactions on Neural Networks, vol. 5, no. 5, 1995, pp. 1061–1070.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1999 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Bachiller, P., Pérez, R.M., Martínez, P., Aguilar, P.L., Díaz, P. (1999). Optimal Hidden Structure for Feedforward Neural Networks. In: Reusch, B. (eds) Computational Intelligence. Fuzzy Days 1999. Lecture Notes in Computer Science, vol 1625. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-48774-3_76
Download citation
DOI: https://doi.org/10.1007/3-540-48774-3_76
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-66050-7
Online ISBN: 978-3-540-48774-6
eBook Packages: Springer Book Archive