Selective Sampling with a Hierarchical Latent Variable Model

  • Hiroshi Mamitsuka
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 2810)


We present a new method which combines a hierarchical stochastic latent variable model and a selective sampling strategy, for learning from co-occurrence events, i.e. a fundamental issue in intelligent data analysis. The hierarchical stochastic latent variable model we employ enables us to use existing background knowledge of observable co-occurrence events as a latent variable. The selective sampling strategy we use iterates selecting plausible non-noise examples from a given data set and running the learning of a component stochastic model alternately and then improves the predictive performance of a component model. Combining the model and the strategy is expected to be effective for enhancing the performance of learning from real-world co-occurrence events. We have empirically tested the performance of our method using a real data set of protein-protein interactions, a typical data set of co-occurrence events. The experimental results have shown that the presented methodology significantly outperformed an existing approach and other machine learning methods compared, and that the presented method is highly effective for unsupervised learning from co-occurrence events.


Unsupervised Learning Latent Variable Model Selective Sampling Protein Class Probabilistic Latent Semantic Analysis 
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.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Hofmann, T.: Unsupervised learning by probabilistic latent semantic analysis. Machine Learning 42, 177–196 (2001)zbMATHCrossRefGoogle Scholar
  2. 2.
    Mamitsuka, H.: Hierarchical latent knowledge analysis for co-occurrence data. In: Proceedings of the Twentieth International Conference on Machine Learning. Morgan Kaufmann, San Francisco (2003)Google Scholar
  3. 3.
    Mamitsuka, H.: Efficient unsupervised mining from noisy data sets. In: Proceedings of the Third SIAM International Conference on Data Mining, SIAM, pp. 239–243 (2003)Google Scholar
  4. 4.
    Dempster, A.P., Laird, N.M., Rubin, D.B.: Maximum likelihood from incomplete data via the EM algorithm. Journal of the Royal Statistical Society: Series B 39, 1–38 (1977)MathSciNetzbMATHGoogle Scholar
  5. 5.
    Bishop, C.M., Tipping, M.E.: A hierarchical latent variable model for data visualization. IEEE Transactions on Pattern Analysis and Machine Intelligence 20, 281–293 (1998)CrossRefGoogle Scholar
  6. 6.
    Schölkopf, B., et al.: Estimating the support of a high-dimensional distribution. Neural Computation 13, 1443–1471 (2001)zbMATHCrossRefGoogle Scholar
  7. 7.
    Joachims, T.: Making large-scale SVM learning practical. In: Schölkopf, B., Burges, C., Smola, A. (eds.) Advances in Kernel Methods – Support Vector Learning. MIT Press, Cambridge (1999)Google Scholar
  8. 8.
    Mewes, H.W., et al.: MIPS: A database for genomes and protein sequences. Nucleic Acids Research 30, 31–34 (2002)CrossRefGoogle Scholar
  9. 9.
    Li, H., Abe, N.: Generalizing case frames using a thesaurus and the MDL principle. Computational Linguistics 24, 217–244 (1998)Google Scholar
  10. 10.
    Rissanen, J.: Modeling by shortest data description. Automatica 14, 465–471 (1978)zbMATHCrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2003

Authors and Affiliations

  • Hiroshi Mamitsuka
    • 1
  1. 1.Institute for Chemical ResearchKyoto UniversityGokasho UjiJapan

Personalised recommendations