Abstract
In applications of artificial neural networks (ANNs), it is common to partition the available data into (at least) two sets. One is then used to train the net, while the other is used as a ‘test set’ to measure the generalization capability of the trained net.
The partition is generally almost completely arbitrary, and little research has been done on the question of what constitutes a good training set, or on how it could be achieved. In this paper, we use a genetic algorithm (GA) to identify a training set for fitting radial basis function (RBF) networks, and test the methodology on two classification problems—one an artificial problem, and the other using real-world data on credit applications for mortgage loans.
In the process, we also exhibit an interesting application of Radcliffe's RAR operator, and present results that suggest the methodology tested here is a viable means of increasing ANN performance.
This is a preview of subscription content, log in via an institution to check access.
Preview
Unable to display preview. Download preview PDF.
References
Reeves, C., Steele, N.: Neural networks for multivariate analysis: results of some cross-validation studies. Proc. of 6 th International Symposium on Applied Stochastic Models and Data Analysis, World Scientific Publishing, Singapore, Vol II (1993), 780–791.
Cook, R., Weisberg, S.: Residuals and Influence in Regression. Chapman and Hall, New York (1982).
Barnett, V., Lewis, T.: Outliers in Statistical Data. Wiley, Chichester (1978).
Dasarathy, B.: Nearest Neighbor (NN) Norms: NN Pattern Classification Techniques. IEEE Computer Society Press, Los Alamitos, CA (1991).
Reeves, C.: Training set selection in neural network applications. In Pearson, D., Albrecht, R., Steele, N.: Proc. of 2nd International Conference on Artificial Neural Nets and Genetic Algorithms. Springer-Verlag, Vienna (1995), 476–478.
Plutowski, M.: Selecting Training Exemplars for Neural Network Learning. PhD Dissertation, University of California, San Diego (1994).
Röbel, A.: The Dynamic Pattern Selection Algorithm: Effective Training and Controlled Generalization of Backpropagation Neural Networks. Technical Report, Technical University of Berlin (1994).
Tambouratzis, T., Tambouratzis, D.: Optimal training pattern selection using a cluster-generating artificial neural network. In Pearson, D., Albrecht, R., Steele, N.: Proc. of 2nd International Conference on Artificial Neural Nets and Genetic Algorithms. Springer-Verlag, Vienna (1995), 472–475.
Radcliffe, N.: Genetic set recombination and its application to neural network topology optimisation. Neural Computing and Applications, 1 (1993), 67–90.
Radcliffe, N., George, F.: A study in set recombination. In Forrest, S. (Ed.) Proceedings of 5 th International Conference on Genetic Algorithms. Morgan Kaufmann, San Mateo, CA (1993), 23–30.
Moody, J., Darken, C.: Fast learning in networks of locally-tuned processing units. Neural Computation, 1 (1990), 281–294.
Bishop, C.: Neural Networks for Pattern Recognition. Clarendon Press, Oxford, UK (1995).
Whitley, L.D., Schaffer, J.D. (Eds.): Proceedings of COGANN-92: International Workshop on Combinations of Genetic Algorithms and Neural Networks. IEEE Computer Society Press, Los Alamitos, CA (1992).
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1998 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Reeves, C.R., Taylor, S.J. (1998). Selection of training data for neural networks by a genetic algorithm. In: Eiben, A.E., Bäck, T., Schoenauer, M., Schwefel, HP. (eds) Parallel Problem Solving from Nature — PPSN V. PPSN 1998. Lecture Notes in Computer Science, vol 1498. Springer, Berlin, Heidelberg. https://doi.org/10.1007/BFb0056905
Download citation
DOI: https://doi.org/10.1007/BFb0056905
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-65078-2
Online ISBN: 978-3-540-49672-4
eBook Packages: Springer Book Archive