Learning Iterative Strategies in Multi-Expert Systems Using SVMs for Digit Recognition

  • Donato Barbuzzi
  • Donato Impedovo
  • Francesco Maurizio Mangini
  • Giuseppe Pirlo
Part of the Lecture Notes in Computer Science book series (LNCS, volume 8156)

Abstract

This paper presents three different learning iterative strategies, in a multi-expert system. In first strategy entire new dataset is used. In second strategy each single classifier selects new samples starting from those on which it performs a misclassification. Finally, the collective behavior of classifiers is studied to select the most profitable samples for knowledge base updating. The experimental results provide a comparison of three approaches under different operating conditions and feedback process. A classifier SVM and four different combination techniques were used by considering the CEDAR (handwritten digit) database. It is shown how results depend by the iterations on the feedback process, as well as by the specific combination decision schema and by data distribution.

Keywords

Feedback-based strategies Instance Selection Multi Expert Systems 

References

  1. 1.
    Kittler, J., Hatef, M., Duin, R.P.W., Matias, J.: On combining classifiers. IEEE Trans. on PAMI 20(3), 226–239 (1998)CrossRefGoogle Scholar
  2. 2.
    Suen, C.Y., Nadal, C., Legault, R., Mai, T.A., Lam, L.: Computer Recognition of unconstrained handwritten numerals. Proc. IEEE 80(7), 1162–1180 (1992)CrossRefGoogle Scholar
  3. 3.
    Liu, C.L., Nakashima, K., Sako, H., Fujisawa, H.: Handwritten digit recognition: benchmarking of state-of-the-art techniques. Pattern Recognition 36(10), 2271–2285 (2003)CrossRefMATHGoogle Scholar
  4. 4.
    Pirlo, G., Impedovo, D.: Fuzzy-Zoning-Based Classification for Handwritten Characters. IEEE Trans. on Fuzzy Systems 19(4), 780–785 (2011)CrossRefGoogle Scholar
  5. 5.
    Suen, C.Y., Tan, J.: Analysis of errors of handwritten digits made by a multitude of classifiers. Pattern Recognition Letters 26(3), 369–379 (2005)CrossRefGoogle Scholar
  6. 6.
    Impedovo, D., Pirlo, G.: Updating Knowledge in Feedback-based Multi-Classifier Systems. In: Proc. of ICDAR, pp. 227–231 (2011)Google Scholar
  7. 7.
    Barbuzzi, D., Impedovo, D., Pirlo, G.: Feedback-Based Strategies In Multi-Expert Systems. In: Sesto Convegno del Gruppo Italiano Ricercatori in Pattern Recognition (2012)Google Scholar
  8. 8.
    Impedovo, D., Pirlo, G., Barbuzzi, D.: Supervised Learning Strategies in Multi-Classifier Systems. In: Proceedings of 11th International Conference on Information Science, Signal Processing and their Applications (ISSPA 2012), pp. 1215–1220 (2012)Google Scholar
  9. 9.
    Barbuzzi, D., Impedovo, D., Pirlo, G.: Benchmarking of Update Learning Strategies on Digit Classifier Systems. In: Proceedings of the 13th International Conference on Frontiers in Handwriting Recognition, pp. 35–40 (2012)Google Scholar
  10. 10.
    Freud, Y., Schapire, R.E.: Decision-theoretic generalization of on-line learning and an application to boosting. J. of Computer and System Sciences 55(1), 119–139 (1997)CrossRefGoogle Scholar
  11. 11.
    Schapire, R.E.: The strength of weak learnability. Machine Learning 5(2), 197–227 (1990)Google Scholar
  12. 12.
    Polikar, R.: Bootstrap-Inspired Techniques in Computational Intelligence. IEEE Signal Processing Magazine 24(4), 59–72 (2007)CrossRefGoogle Scholar
  13. 13.
    Hull, J.: A database for handwritten text recognition research. IEEE T-PAMI 16(5), 550–554 (1994)CrossRefGoogle Scholar
  14. 14.
    Impedovo, S., Modugno, R., Ferrante, A., Pirlo, G.: Zoning Methods for Hand-written Character Recognition: An Overview. In: Proceedings of the 12th International Conference on Frontiers in Handwriting Recognition (ICFHR 2012), Kolkata, India, November 16-18, pp. 329–334 (2010)Google Scholar
  15. 15.
    Impedovo, D., Modugno, R., Pirlo, G.: New Advancements in Zoning-Based Recognition of Handwritten Characters. In: Proc. XIII International Conference on Frontiers in Handwriting Recognition (ICFHR 2012), Monopoli, Bari, Italy, September 18-20, pp. 661–665 (2012)Google Scholar
  16. 16.
    Impedovo, S., et al.: Feature Membership Functions in Voronoi-Based Zoning. In: Serra, R., Cucchiara, R. (eds.) AI*IA 2009. LNCS, vol. 5883, pp. 202–211. Springer, Heidelberg (2009)CrossRefGoogle Scholar
  17. 17.
    Impedovo, S., Modugno, R., Pirlo, G.: Analysis of Membership Functions for Voronoi-based Classification. In: Proceedings of the 12th Interational Conference on Frontiers in Handwriting Recognition (ICFHR 2012), November 16-18, pp. 220–225. IEEE Computer Society Press, Kolkata (2010)CrossRefGoogle Scholar
  18. 18.
    Pirlo, G., Impedovo, D.: Adaptive Membership Functions for Hand-Written Character Recognition by Voronoi-based Image Zoning. IEEE Transactions on Image Processing 21(9), 3827–3837 (2012)MathSciNetCrossRefGoogle Scholar
  19. 19.
    Impedovo, S., Pirlo, G.: Tuning between Exponential Functions and Zones for Membership Functions Selection in Voronoi-based Zoning for Handwritten Character Recognition. In: Proc. of the 11th International Conference on Document Analysis and Recognition (ICDAR 2011), September 18-21, pp. 997–1001. IEEE Computer Society, Beijing (2011) ISBN: 978-0-7695-4520-2CrossRefGoogle Scholar
  20. 20.
    Impedovo, D., Modugno, R., Pirlo, G.: Score Normalization by Dynamic Time Warping. In: Proceedings of the International Conference on Computational Intelligence for Measurement Systems and Applications (CIMSA), Taranto, Italy, September 6-8, pp. 82–85. IEEE Computer Society Press, Taranto (2010) ISBN: 978-1-4244-7229-1Google Scholar
  21. 21.
    Pirlo, G., Impedovo, D.: Adaptive Score Normalization for Multi-Classifier Systems. IEEE Signal Processing Letters 19(12), 837–840 (2012) ISSN: 1070-9908 CrossRefGoogle Scholar
  22. 22.
    Impedovo, D., Pirlo, G., Sarcinella, L., Stasolla, E.: Artificial Classifier Generation for Multi-Expert System Evaluation. In: Proceedings of the 12th International Conference on Frontiers in Handwriting Recognition (ICFHR 2012), November 16-18, pp. 42–426. IEEE Computer Society Press, Kolkata (2010) ISBN: 978-0-7695-4221-8Google Scholar
  23. 23.
    Bovino, L., Dimauro, G., Impedovo, S., Lucchese, M.G., Modugno, R., Pirlo, G., Salzo, A., Sarcinella, L.: On the Combination of Abstract-Level Classifiers. International Journal on Document Analysis and Recognition 6, 42–54 (2003) ISSN 1433-2833CrossRefGoogle Scholar
  24. 24.
    Frinken, V., Bunke, H.: Evaluating Retraining Rules for Semi-Supervised Learning in Neural Network Based Cursive Word Recognition. In: Proc. of ICDAR, pp. 31–35 (2009)Google Scholar
  25. 25.
    Frinken, V., Fischer, A., Bunke, H., Fornes, A.: Co-Training for Handwritten Word Recognition. In: Proc. of ICDAR, pp. 314–318 (2011)Google Scholar
  26. 26.
    Blum, A., Mitchell, T.: Combining Labeled and Unlabeled Data with Co-Training. In: ACM Proc. of COLT, pp. 92–100 (1998)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2013

Authors and Affiliations

  • Donato Barbuzzi
    • 1
  • Donato Impedovo
    • 2
  • Francesco Maurizio Mangini
    • 1
  • Giuseppe Pirlo
    • 1
  1. 1.Department of Computer ScienceUniversity of BariBariItaly
  2. 2.Department of Electrical and Electronic EngineeringPolytechnic of BariBariItaly

Personalised recommendations