Multiple-Instance Learning Via Random Walk

  • Dong Wang
  • Jianmin Li
  • Bo Zhang
Part of the Lecture Notes in Computer Science book series (LNCS, volume 4212)


This paper presents a decoupled two stage solution to the multiple-instance learning (MIL) problem. With a constructed affinity matrix to reflect the instance relations, a modified Random Walk on a Graph process is applied to infer the positive instances in each positive bag. This process has both a closed form solution and an efficient iterative one. Combined with the Support Vector Machine (SVM) classifier, this algorithm decouples the inferring and training stages and converts MIL into a supervised learning problem. Compared with previous algorithms on several benchmark data sets, the proposed algorithm is quite competitive in both computational efficiency and classification accuracy.


Support Vector Machine Positive Instance Negative Instance Multiple Instance Learning Support Vector Machine Parameter 


  1. 1.
    Dietterich, T.G., Lathrop, R.H., Lozano-Perez, T.: Solving the multiple instance problem with axis-parallel rectangles. Artificial Intelligence 89, 31–71 (1997)MATHCrossRefGoogle Scholar
  2. 2.
    Maron, O.: Learning from Ambiguity. PhD thesis. MIT (1998)Google Scholar
  3. 3.
    Zhang, Q., Goldman, S.: Em-dd: An improved multiple-instance learning technique. In: NIPS, vol. 14, pp. 1073–1080 (2002)Google Scholar
  4. 4.
    Andrews, S., Tsochantaridis, I., Hofmann, T.: Support vector machines for multiple-instance learning. In: NIPS, vol. 15 (2003)Google Scholar
  5. 5.
    Chen, Y., Wang, J.Z.: Image categorization by learning and reasoning with regions. Journal of Machine Learning Research 5, 913–939 (2004)Google Scholar
  6. 6.
    Ray, S., Craven, M.: Supervised versus multiple instance learning: An empirical comparison. In: Proc. 22th International Conf. on Machine Learning (2005)Google Scholar
  7. 7.
    Ramon, J., Raedt, L.D.: Multi instance neural networks. In: Proceedings of ICML 2000, Workshop on Attribute-Value and Relational Learning (2000)Google Scholar
  8. 8.
    Wang, J., Zucker, J.D.: Solving the multiple-instance problem: a lazy learning approach. In: Proc. 17th Int. Conf. on Machine Learning, pp. 1119–1125 (2000)Google Scholar
  9. 9.
    Ruffo, G.: Learning single and multiple instance decision trees for computer security applications. PhD thesis, Universitá di Torino, Torino, Italy (2000)Google Scholar
  10. 10.
    Gärtner, T., Flach, P.A., Kowalczyk, A., Smola, A.J.: Multi-instance kernels. In: Proc. 19th International Conf. on Machine Learning, pp. 179–186 (2002)Google Scholar
  11. 11.
    Tao, Q., Scott, S.D., Vinodchandran, N.V.: Svm-based generalized multiple-instance learning via approximate box counting. In: Proc. 21th International Conf. on Machine Learning (2004)Google Scholar
  12. 12.
    Auer, P., Ortner, R.: A boosting approach to multiple instance learning. In: Proc. of the 15th European Conf. on Machine Learning (2004)Google Scholar
  13. 13.
    Rahmani, R., Goldman, S.A., Zhang, H., Krettek, J., Fritts, J.E.: Localized content based image retrieval. In: Proc. ACM Int. Conf. on Multimedia (ACM MM) (2005)Google Scholar
  14. 14.
    Brin, S., Page, L.: The anatomy of a large-scale hypertextual web search engine. In: Proc. 7th Int. World Wide Web Conf. (1998)Google Scholar
  15. 15.
    Zhou, D., Bousquet, O., Lal, T.N., Weston, J., Schölkopf, B.: Learning with local and global consistency. In: NIPS, vol. 15, pp. 237–244 (2003)Google Scholar
  16. 16.
    Szummer, M., Jaakkola, T.: Partially labeled classification with markov random walks. In: NIPS (2002)Google Scholar
  17. 17.
    Zhu, X., Ghahramani, Z., Lafferty, J.: Semi-supervised learning using gaussian fields and harmonic functions. In: ICML (2003)Google Scholar
  18. 18.
    Belkin, M., Niyogi, P.: Using manifold stucture for partially labeled classification. In: NIPS (2003)Google Scholar
  19. 19.
    Zhou, D., Huang, J., Schölkopf, B.: Learning from labeled and unlabeled data on a directed graph. In: Proc. 22th International Conf. on Machine Learning (2005)Google Scholar
  20. 20.
    Zhu, X., Lafferty, J.: Harmonic mixtures: combining mixture models and graph-based methods for inductive and scalable semi-supervised learning. In: Proc. 22th International Conf. on Machine Learning (2005)Google Scholar
  21. 21.
    Kulis, B., Basu, S., Dhillon, I.S., Mooney, R.J.: Semi-supervised graph clustering: A kernel approach. In: Proc. 22th International Conf. on Machine Learning (2005)Google Scholar
  22. 22.
    Breitenbach, M., Grudic, G.Z.: Clustering through ranking on manifolds. In: Proc. 22th International Conf. on Machine Learning (2005)Google Scholar
  23. 23.
    Seneta, E.: Non-Negative Matrices and Markov Chains. Springer, Heidelberg (1981)MATHGoogle Scholar
  24. 24.
    Chang, C.C., Lin, C.J.: LIBSVM: a library for support vector machines (2001)Google Scholar
  25. 25.
    Zhou, D., Bousquet, O., Lal, T.N., Weston, J., Schölkopf, B.: Ranking on data manifolds. In: NIPS, vol. 15, pp. 169–176 (2003)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • Dong Wang
    • 1
  • Jianmin Li
    • 1
  • Bo Zhang
    • 1
  1. 1.State Key Laboratory of Intelligent Technology and System, Department of Computer Science and TechnologyTsinghua UniversityBeijingP.R. China

Personalised recommendations