Abstract
An idea of training multilayer perceptron (MLP) with multiple classifier systems is introduced in this paper. Instead of crisp class membership i.e. {0, 1}, the desired output of training samples is assigned by multiple classifier systems. Trained with these samples, the network is more reliable and processes better outlier rejection ability. The effectiveness of this idea is confirmed by a series of experiments based on bank check handwritten numeral recognition. Experimental results show that for some recognition applications where high reliability is needed, MLP trained with multiple classifier systems label samples is more qualified than that of MLP trained with crisp label samples.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Cho, S.B.: Neural-network Classifiers for Recognizing Totally Unconstrained Handwritten Numerals. IEEE Trans. Neural Networks 3(1), 43–53 (1997)
Cun, Y., Le., M.O., Boser, B., Denker, J.S., Henderson, D., Howard, R.E., Hubbard, W., Jacket, L.D., Baird, H.S.: Handwritten Zip Code Recognition with Multilayer Networks. In: Proceedings of the 10th International Conference on Pattern Recognition, vol. 2, pp. 35–40 (1990)
Wu, X.J., Suen, C.Y.: Injection of Human Knowledge into The Rejection Criterion of a Neural Network Classifier. In: Joint 9th IFSA World Congress and 20th NAFIPS International Conference, vol. 1, pp. 499–505 (2001)
Gori, M., Scarselli, F.: Are Multilayer Perceptrons Adequate for Pattern Recognition and Verification? IEEE Trans. Pattern Anal. and Mach. Intell. 20(11), 1121–1132 (1998)
Lippmann, R.P.: Pattern Classification Using Neural Networks. IEEE Communications Magazine 27(11), 47–50, 59-64 (1989)
Chiang, J.-H.: A Hybrid Neural Network Model in Handwritten Word Recognition. Neural Networks 11(2), 337–346 (1998)
Gader, P., Mohamed, M., Chiang, J.-H.: Comparison of Crisp and Fuzzy Character Neural Networks in Handwritten Word Recognition. IEEE Trans. Fuzzy Systems 3(3), 357–363 (1995)
Huang, Y.S., Suen, C.Y.: A Method of Combining Multiple Classifiers-A Neural Network Approach. In: Proceedings of the 12th IAPR International Conference on Pattern Recognition- Conference B: Computer Vision & Image Processing, vol. 2, pp. 9–13, 473-475 (1994)
Altınçay, H., Demirekler, M.: Undesirable Effects of Output Normalization in Multiple Classifier Systems. Pattern Recognition Letters 24, 1163–1170 (2003)
Duin, R.P.W., Tax, D.M.J.: Classifier Conditional Posterior Probabilities. In: Amin, A., Pudil, P., Dori, D. (eds.) SPR 1998 and SSPR 1998. LNCS, vol. 1451, pp. 611–619. Springer, Heidelberg (1998)
Kuncheva, L.I., Bezdek, J.C., Duin, R.P.W.: Decision Templates for Multiple Classifier Fusion: An Experimental Comparison. Pattern Recognition 34, 299–314 (2001)
Liu, C.-L., Sako, H., Fujisawa, H.: Performance Evaluation of Pattern Classifiers for Handwritten Character Recogniton. International Journal on Document Analysis and Recognition 4, 191–204 (2002)
Kato, N., Suzuki, M., Omachi, S., Aso, H., Nemoto, Y.: A Handwritten Character Recognition System Using Directional Element Feature And Asymmetric Mahalanobis Distance. IEEE Trans. Pattern Anal. and Mach. Intell. 21(3), 258–262 (1999)
Gader, P.D., Mohamed, M., Chiang, J.-H.: Handwritten Word Recognition with Character and Inter-Character Neural Networks. IEEE Trans. Systems, Man and Cybernetics-Part B 27(1), 158–164 (1997)
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
Zhu, H., Liu, J., Tang, X., Huang, J. (2004). Training Multilayer Perceptron with Multiple Classifier Systems. In: Yin, FL., Wang, J., Guo, C. (eds) Advances in Neural Networks – ISNN 2004. ISNN 2004. Lecture Notes in Computer Science, vol 3173. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-28647-9_147
Download citation
DOI: https://doi.org/10.1007/978-3-540-28647-9_147
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-22841-7
Online ISBN: 978-3-540-28647-9
eBook Packages: Springer Book Archive