Multiple Classifier Systems for Adversarial Classification Tasks

  • Battista Biggio
  • Giorgio Fumera
  • Fabio Roli
Part of the Lecture Notes in Computer Science book series (LNCS, volume 5519)


Pattern classification systems are currently used in security applications like intrusion detection in computer networks, spam filtering and biometric identity recognition. These are adversarial classification problems, since the classifier faces an intelligent adversary who adaptively modifies patterns (e.g., spam e-mails) to evade it. In these tasks the goal of a classifier is to attain both a high classification accuracy and a high hardness of evasion, but this issue has not been deeply investigated yet in the literature. We address it under the viewpoint of the choice of the architecture of a multiple classifier system. We propose a measure of the hardness of evasion of a classifier architecture, and give an analytical evaluation and comparison of an individual classifier and a classifier ensemble architecture. We finally report an experimental evaluation on a spam filtering task.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Globerson, A., Roweis, S.T.: Nightmare at test time: robust learning by feature deletion. In: Cohen, W.W., Moore, A. (eds.) ICML. ACM International Conference Proceeding Series, vol. 148, pp. 353–360. ACM, New York (2006)CrossRefGoogle Scholar
  2. 2.
    Perdisci, R., Gu, G., Lee, W.: Using an ensemble of one-class svm classifiers to harden payload-based anomaly detection systems. In: International Conference on Data Mining (ICDM), pp. 488–498. IEEE Computer Society, Los Alamitos (2006)CrossRefGoogle Scholar
  3. 3.
    Jorgensen, Z., Zhou, Y., Inge, M.: A multiple instance learning strategy for combating good word attacks on spam filters. Journal of Machine Learning Research 9, 1115–1146 (2008)Google Scholar
  4. 4.
    Lowd, D., Meek, C.: Adversarial learning. In: Press, A. (ed.) Proceedings of the Eleventh ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD), Chicago, IL (2005)Google Scholar
  5. 5.
    Dalvi, N., Domingos, P., Mausam, S.S., Verma, D.: Adversarial classification. In: Tenth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD), Seattle, pp. 99–108 (2004)Google Scholar
  6. 6.
    Barreno, M., Nelson, B., Sears, R., Joseph, A.D., Tygar, J.D.: Can machine learning be secure? In: ASIACCS 2006: Proceedings of the 2006 ACM Symposium on Information, computer and communications security, pp. 16–25. ACM, New York (2006)CrossRefGoogle Scholar
  7. 7.
    Kittler, J., Hatef, M., Duin, R.P., Matas, J.: On combining classifiers. IEEE Transactions on Pattern Analysis and Machine Intelligence 20(3), 226–239 (1998)CrossRefGoogle Scholar
  8. 8.
    Ross, A.A., Nandakumar, K., Jain, A.K.: Handbook of Multibiometrics. Springer Publishers, Heidelberg (2006)Google Scholar
  9. 9.
    Haindl, M., Kittler, J., Roli, F. (eds.): MCS 2007. LNCS, vol. 4472. Springer, Heidelberg (2007)zbMATHGoogle Scholar
  10. 10.
    Giacinto, G., Roli, F., Didaci, L.: Fusion of multiple classifiers for intrusion detection in computer networks. Pattern Recognition Letters 24, 1795–1803 (2003)CrossRefzbMATHGoogle Scholar
  11. 11.
    Biggio, B., Fumera, G., Roli, F.: Evade hard multiple classifier systems. In: Okun, O., Valentini, G. (eds.) Supervised and Unsupervised Ensemble Methods and Their Applications. Studies in Computational Intelligence. Springer, Heidelberg (2009) (in press)Google Scholar
  12. 12.
    Chang, C.C., Lin, C.J.: LIBSVM: a library for support vector machines (2001),

Copyright information

© Springer-Verlag Berlin Heidelberg 2009

Authors and Affiliations

  • Battista Biggio
    • 1
  • Giorgio Fumera
    • 1
  • Fabio Roli
    • 1
  1. 1.Dept. of Electrical and Electronic Eng.Univ. of CagliariCagliariItaly

Personalised recommendations