A New Constructive Algorithm for Designing and Training Artificial Neural Networks
This paper presents a new constructive algorithm, called problem dependent constructive algorithm (PDCA), for designing and training artificial neural networks (ANNs). Unlike most previous studies, PDCA puts emphasis on architectural adaptation as well as function level adaptation. The architectural adaptation is done by determining automatically the number of hidden layers in an ANN and of neurons in hidden layers. The function level adaptation, is done by training each hidden neuron with a different training set. PDCA uses a constructive approach to achieve both the architectural as well as function level adaptation. It has been tested on a number of benchmark classification problems in machine learning and ANNs. The experimental results show that PDCA can produce ANNs with good generalization ability in comparison with other algorithms.
KeywordsArtificial neural networks (ANNs) architectural adaptation function level adaptation constructive approach and generalization ability
Unable to display preview. Download preview PDF.
- 1.Fahlman, S.E., Lebiere, C.: The cascade-correlation learning architecture. In: Touretzky, D.S. (ed.) Advances in Neural Information Processing System, vol. 2, pp. 524–532. Morgan Kaufmann, San Mateo, CA (1990)Google Scholar
- 2.Fahlman, S.E.: An empirical study of learning speed in back-propagation networks. Tech Report CMU-CS-88-162, Carnegie Mellon University (1988)Google Scholar
- 9.Monirul Islam, Md., Murase, K.: A new algorithm to design compact two-hidden-layer artificial neural networks. Neural Networks 14, 1265–1278 (2001)Google Scholar
- 13.Prechelt, L.: PROBEN1- A set of neural network benchmark problems and benchmarking rules. Technical Report 21/94, Faculty of Informatics, University of Karlsruhe, Germany (1994)Google Scholar
- 14.Riedmiller, M., Braun, H.: RPROP - A Fast Adaptive Learning Algorithm. In: Proc. of the 1992 International Symposium on Computer and Information Sciences, Turkey, November 1992, pp. 279–285 (1992)Google Scholar
- 15.Schapire, R.E.: The strength of weak learnability. Machine Learning 5, 197–227 (1990)Google Scholar