Improving SVM Training by Means of NTIL When the Data Sets Are Imbalanced

  • Carlos E. Vivaracho
Part of the Lecture Notes in Computer Science book series (LNCS, volume 4203)


This paper deals with the problem of training a discriminative classifier when the data sets are imbalanced. More specifically, this work is concerned with the problem of classify a sample as belonging, or not, to a Target Class (TC), when the number of examples from the “Non-Target Class” (NTC) is much higher than those of the TC. The effectiveness of the heuristic method called Non Target Incremental Learning (NTIL) in the task of extracting, from the pool of NTC representatives, the most discriminant training subset with regard to the TC, has been proved when an Artificial Neural Network is used as classifier (ISMIS 2003). In this paper the effectiveness of this method is also shown for Support Vector Machines.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Batista, G.: A study of the behavior of several methods for balancing machine learning training data. SIGKDD explorations 6(1), 20–29 (2004)CrossRefMathSciNetGoogle Scholar
  2. 2.
    Chawla, N.V., Japkowicz, N., Kotcz, A.: Editorial: Special issue on learning from imbalance data sets. SIGKDD explorations 6(1), 1–6 (2004)CrossRefGoogle Scholar
  3. 3.
    Cortes, C., Vapnik, V.: Support-vector network. Machine Learning (20), 273–297 (1995)Google Scholar
  4. 4.
    Farrel, K.R., Mammone, R.J., Assaleh, K.T.: Speaker recognition using neural networks and conventional classifiers. IEEE Transations on Speech and Audio Processing, part II, 2(1) (1994)Google Scholar
  5. 5.
    Japkowicz, N., Stephen, S.: The class imbalance problem: A sistematic study. Intelligent Data Analysis 6(5), 429–449 (2002)zbMATHGoogle Scholar
  6. 6.
    Juszczak, P., Duin, R.P.W.: Uncertainty sampling methods for one-class classifiers. In: Proc. of the Workshop on Learning from Imbalanced Datasets II, ICML (2003)Google Scholar
  7. 7.
    Mansfield, A.J., Wayman, J.L.: Best pratices in testing and reporting performance of biometric devices. version 2.01. Technical report (2002)Google Scholar
  8. 8.
    del Brio, B.M., Sanz Molina, A.: Redes Neuronales y Sistemas Borrosos. Ra-Ma (1997)Google Scholar
  9. 9.
    Osuna, E., Freund, R., Girosi, F.: Support vector machines: Training and applications. Technical report (1997)Google Scholar
  10. 10.
    Solomonoff, A., Quillen, C., Campbell, W.M.: Channel compensation for svm speaker recognition. In: Proc. Odyssey 2004, the Speaker and Language Recognition Workshop, May 31 - June 3 (2004)Google Scholar
  11. 11.
    Vapnik, V.: Statistical Learning Theory. Wiley, New York (1998)zbMATHGoogle Scholar
  12. 12.
    Vivaracho, C.E., Ortega-Garcia, J., Alonso, L., Moro, Q.I.: Extracting the most discriminant subset from a pool of candidates to optimize discriminant classifier training. In: Zhong, N., Raś, Z.W., Tsumoto, S., Suzuki, E. (eds.) ISMIS 2003. LNCS (LNAI), vol. 2871, pp. 640–645. Springer, Heidelberg (2003)CrossRefGoogle Scholar
  13. 13.
    Vivaracho-Pascual, C., Ortega-Garcia, J., Alonso-Romero, L., Moro-Sancho, Q.: A comparative study of mlp-bsed artificial neural networks in text-indenpendent speaker verification against gmm-based systems. In: Dalsgaard, B.L.P., Benner, H. (eds.) Proc. of Eurospeech 2001, ISCA September 3-7, vol. 3, pp. 1753–1756 (2001)Google Scholar
  14. 14.
    Weiss, G.M.: Mining with rarity: A unifing framework. SIGKDD explorations 6(1), 7–19 (2004)CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • Carlos E. Vivaracho
    • 1
  1. 1.Dep. InformáticaU. de ValladolidSpain

Personalised recommendations