Class-Dependent Dissimilarity Measures for Multiple Instance Learning

  • Veronika Cheplygina
  • David M. J. Tax
  • Marco Loog
Part of the Lecture Notes in Computer Science book series (LNCS, volume 7626)

Abstract

Multiple Instance Learning (MIL) is concerned with learning from sets (bags) of feature vectors (instances), where the individual instance labels are ambiguous. In MIL it is often assumed that positive bags contain at least one instance from a so-called concept in instance space, whereas negative bags only contain negative instances. The classes in a MIL problem are therefore not treated in the same manner. One of the ways to classify bags in MIL problems is through the use of bag dissimilarity measures. In current dissimilarity approaches, such dissimilarity measures act on the bag as a whole and do not distinguish between positive and negative bags. In this paper we explore whether this is a reasonable approach and when and why a dissimilarity measure that is dependent on the bag label, might be more appropriate.

Keywords

Dissimilarity Measure Positive Instance Inductive Logic Programming Multiple Instance Learn Prototype Selection 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

References

  1. 1.
    Andrews, S., Hofmann, T., Tsochantaridis, I.: Multiple instance learning with generalized support vector machines. In: Proc. of the National Conference on Artificial Intelligence, pp. 943–944. AAAI Press, MIT Press, Menlo Park, Cambridge (2002)Google Scholar
  2. 2.
    Bailey, A.: Class-dependent features and multicategory classification. Ph.D. thesis, Citeseer (2001)Google Scholar
  3. 3.
    Chen, Y., Bi, J., Wang, J.: Miles: Multiple-instance learning via embedded instance selection. IEEE Transactions on Pattern Analysis and Machine Intelligence 28(12), 1931–1947 (2006)CrossRefGoogle Scholar
  4. 4.
    Cheplygina, V., Tax, D., Loog, M.: Does one rotten apple spoil the whole barrel? In: International Conference on Pattern Recognition (in press)Google Scholar
  5. 5.
    De Wachter, M., Demuynck, K., Wambacq, P., Van Compernolle, D.: A locally weighted distance measure for example based speech recognition. In: International Conference on Acoustics, Speech, and Signal Processing, vol. 1, pp. I–181. IEEE (2004)Google Scholar
  6. 6.
    Dietterich, T., Lathrop, R., Lozano-Pérez, T.: Solving the multiple instance problem with axis-parallel rectangles. Artificial Intelligence 89(1-2), 31–71 (1997)MATHCrossRefGoogle Scholar
  7. 7.
    Gärtner, T., Flach, P., Kowalczyk, A., Smola, A.: Multi-instance kernels. In: Proc. of the 19th Int. Conf. on Machine Learning, pp. 179–186 (2002)Google Scholar
  8. 8.
    Kummamuru, K., Krishnapuram, R., Agrawal, R.: On learning asymmetric dissimilarity measures. In: International Conference on Data Mining, p. 4. IEEE (2005)Google Scholar
  9. 9.
    Maron, O., Lozano-Pérez, T.: A framework for multiple-instance learning. In: Advances in Neural Information Processing Systems, pp. 570–576. Morgan Kaufmann Publishers (1998)Google Scholar
  10. 10.
    Paredes, R., Vidal, E.: A class-dependent weighted dissimilarity measure for nearest neighbor classification problems. Pattern Recognition Letters 21(12), 1027–1036 (2000)MATHCrossRefGoogle Scholar
  11. 11.
    Pękalska, E., Duin, R.P.W.: The dissimilarity representation for pattern recognition: foundations and applications, vol. 64. World Scientific Pub. Co. Inc. (2005)Google Scholar
  12. 12.
    Pękalska, E., Duin, R., Paclík, P.: Prototype selection for dissimilarity-based classifiers. Pattern Recognition 39(2), 189–208 (2006)MATHCrossRefGoogle Scholar
  13. 13.
    Rahmani, R., Goldman, S., Zhang, H., Krettek, J., Fritts, J.: Localized content based image retrieval. In: Proc. of the 7th ACM SIGMM International Workshop on Multimedia Information Retrieval, pp. 227–236. ACM (2005)Google Scholar
  14. 14.
    Sørensen, L., Loog, M., Tax, D., Lee, W., de Bruijne, M., Duin, R.: Dissimilarity-Based Multiple Instance Learning. In: Hancock, E.R., Wilson, R.C., Windeatt, T., Ulusoy, I., Escolano, F. (eds.) SSPR&SPR 2010. LNCS, vol. 6218, pp. 129–138. Springer, Heidelberg (2010)CrossRefGoogle Scholar
  15. 15.
    Srinivasan, A., Muggleton, S., King, R.: Comparing the use of background knowledge by inductive logic programming systems. In: Proceedings of the 5th International Workshop on Inductive Logic Programming, pp. 199–230 (1995)Google Scholar
  16. 16.
    Tao, Q., Scott, S., Vinodchandran, N., Osugi, T.: Svm-based generalized multiple-instance learning via approximate box counting. In: Proc. of the 21st Int. Conf. on Machine learning, p. 101. ACM (2004)Google Scholar
  17. 17.
    Tax, D., Loog, M., Duin, R., Cheplygina, V., Lee, W.: Bag dissimilarities for multiple instance learning. Similarity-Based Pattern Recognition, 222–234 (2011)Google Scholar
  18. 18.
    Wang, J.: Solving the multiple-instance problem: A lazy learning approach. In: Proc. of the 17th Int. Conf. on Machine Learning (2000)Google Scholar
  19. 19.
    Weinberger, K., Saul, L.: Distance metric learning for large margin nearest neighbor classification. The Journal of Machine Learning Research 10, 207–244 (2009)MATHGoogle Scholar
  20. 20.
    Zafra, A., Pechenizkiy, M., Ventura, S.: Reducing dimensionality in multiple instance learning with a filter method. Hybrid Artificial Intelligence Systems, 35–44 (2010)Google Scholar
  21. 21.
    Zhao, C., Shi, W., Deng, Y.: A new hausdorff distance for image matching. Pattern Recognition Letters 26(5), 581–586 (2005)CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2012

Authors and Affiliations

  • Veronika Cheplygina
    • 1
  • David M. J. Tax
    • 1
  • Marco Loog
    • 1
  1. 1.Pattern Recognition LaboratoryDelft University of TechnologyThe Netherlands

Personalised recommendations