Abstract
In this paper, a new learning method is proposed to build Support Vector Machines (SVMs) Binary Decision Functions (BDF) of reduced complexity and efficient generalization. The aim is to build a fast and efficient SVM classifier. A criterion is defined to evaluate the Decision Function Quality (DFQ) which blendes recognition rate and complexity of a BDF. Vector Quantization (VQ) is used to simplify the training set. A model selection based on the selection of the simplification level, of a feature subset and of SVM hyperparameters is performed to optimize the DFQ. Search space for selecting the best model being huge, Tabu Search (TS) is used to find a good sub-optimal model on tractable times. Experimental results show the efficiency of the method.
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
Preview
Unable to display preview. Download preview PDF.
References
Platt, J.: Fast training of SVMs using sequential minimal optimization, Advances in kernel methods-support vector learning, pp. 185–208. MIT Press, Cambridge (1999)
Yu, H., Yang, J., Han, J.: Classifying large data sets using SVM with hierarchical clusters. In: SIGKDD, pp. 306–315 (2003)
Lebrun, G., Charrier, C., Cardot, H.: SVM training time reduction using vector quantization. In: ICPR, vol. 1, pp. 160–163 (2004)
Chang, C.C., Lin, C.J.: Libsvm: a library for support vector machines. Sofware Available at (2001), http://www.csie.ntu.edu.tw/~cjlin/libsvm
Ou, Y.Y., Chen, C.Y., Hwang, S.C., Oyang, Y.J.: Expediting model selection for SVMs based on data reduction. In: IEEE Proc. SMC, pp. 786–791 (2003)
Tsang, I.W., Kwok, J.T., Cheung, P.M.: Core vector machines: Fast SVM training on very large data sets. In: JMLR, vol. 6, pp. 363–392 (2005)
Lebrun, G., Charrier, C., Lezoray, O., Meurie, C., Cardot, H.: Fast pixel classification by SVM using vector quantization, tabu search and hybrid color space. In: CAIP, pp. 685–692 (2005)
Chapelle, O., Vapnik, V., Bousquet, O., Mukherjee, S.: Choosing multiple parameters for support vector machines. Machine Learning 46, 131–159 (2002)
Chapelle, O., Vapnik, V.: Model selection for support vector machines. Advances in Neural Information Processing Systems 12, 230–236 (1999)
Fröhlich, H., Chapelle, O., Schölkopf, B.: Feature selection for support vector machines using genetic algorithms. IJAIT 13, 791–800 (2004)
Rifkin, R., Klautau, A.: In defense of one-vs-all classification. JMLR 5, 101–141 (2004)
Christianini, N.: Dimension reduction in text classification with support vector machines. In: JMLR, vol. 6, pp. 37–53 (2005)
Gersho, A., Gray, R.M.: Vector Quantization and Signal Compression. Kluwer Academic, Dordrecht (1991)
Staelin, C.: Parameter selection for support vector machines (2002), http://www.hpl.hp.com/techreports/2002/HPL-2002-354R1.html
Glover, F., Laguna, M.: Tabu search. Kluwer Academic Publishers, Dordrecht (1997)
Korycinski, D., Crawford, M.M., Barnes, J.W.: Adaptive feature selection for hyperspectral data analysis. In: SPIE, vol. 5238, pp. 213–225 (2004)
Vapnik, V.N.: Statistical Learning Theory. Wiley edn., New York (1998)
Blake, C., Merz, C.: Uci repository of machine learning databases. In: Advances in kernel methods, support vector learning (1998)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2006 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Lebrun, G., Lezoray, O., Charrier, C., Cardot, H. (2006). A New Model Selection Method for SVM. In: Corchado, E., Yin, H., Botti, V., Fyfe, C. (eds) Intelligent Data Engineering and Automated Learning – IDEAL 2006. IDEAL 2006. Lecture Notes in Computer Science, vol 4224. Springer, Berlin, Heidelberg. https://doi.org/10.1007/11875581_12
Download citation
DOI: https://doi.org/10.1007/11875581_12
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-45485-4
Online ISBN: 978-3-540-45487-8
eBook Packages: Computer ScienceComputer Science (R0)