Towards Fewer Seeds for Network Discovery

Conference paper
Part of the Advances in Intelligent Systems and Computing book series (AISC, volume 360)

Abstract

In most machine-learning problems, unlabeled data used in conjunction with a small amount of labeled data, can result in considerable improvement in learning accuracy. The goal of semi-supervised learning is to learn from these initial labeled data to predict class labels accurately. An important optimization question is to select these initial labeled instances efficiently. In this paper, we explore this problem. We propose two algorithms: one based on clustering and another based on random walk on the network. We consider four important criteria for selecting a data point as seed, with the general aim of choosing data points which are a good summary of the network. We show experimental results on different network datasets and show improvements over a recent result.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Rajan, S., Yankov, D., Gaffney, S.J., Ratnaparkhi, A.: A large-scale active learning system for topical categorization on the web. In: Proceedings of the 19th International Conference on World Wide Web, WWW 2010, pp. 791–800. ACM, New York (2010)Google Scholar
  2. 2.
    Shi, L., Zhao, Y., Tang, J.: Combining link and content for collective active learning. In: Proceedings of the 19th ACM International Conference on Information and Knowledge Management, CIKM 2010, pp. 1829–1832. ACM, New York (2010)Google Scholar
  3. 3.
    Zhu, X., Lafferty, J., Ghahramani, Z.: Combining active learning and semi-supervised learning using gaussian fields and harmonic functions. In: ICML 2003 Workshop on The Continuum from Labeled to Unlabeled Data in Machine Learning and Data Mining, pp. 58–65 (2003)Google Scholar
  4. 4.
    Ehara, Y., Sato, I., Oiwa, H., Nakagawa, H.: Understanding seed selection in bootstrapping. In: Proceedings of TextGraphs-8 Graph-based Methods for Natural Language Processing, pp. 44–52. Association for Computational Linguistics, Seattle (2013)Google Scholar
  5. 5.
    Lin, F., Cohen, W.W.: Semi-supervised classification of network data using very few labels. In: Proceedings of the 2010 International Conference on Advances in Social Networks Analysis and Mining, ASONAM 2010, pp. 192–199. IEEE Computer Society, Washington, DC (2010)CrossRefGoogle Scholar
  6. 6.
    Tong, H., Faloutsos, C., Pan, J.Y.: Fast random walk with restart and its applications. In: Proceedings of the Sixth International Conference on Data Mining, ICDM 2006, pp. 613–622. IEEE Computer Society, Washington, DC (2006)Google Scholar
  7. 7.
    Katz, L.: A new status index derived from sociometric analysis. Psychometrika 18, 39–43 (1953)CrossRefMATHGoogle Scholar
  8. 8.
    Sen, P., Namata, G.M., Bilgic, M., Getoor, L., Gallagher, B., Eliassi-Rad, T.: Collective classification in network data. AI Magazine 29, 93–106 (2008)Google Scholar

Copyright information

© Springer International Publishing Switzerland 2015

Authors and Affiliations

  1. 1.Max Planck Institute for InformaticsSaarbrückenGermany

Personalised recommendations