L2-SVM Training with Distributed Data

  • Stefano Lodi
  • Ricardo Ñanculef
  • Claudio Sartori
Part of the Lecture Notes in Computer Science book series (LNCS, volume 5774)

Abstract

We propose an algorithm for the problem of training a SVM model when the set of training examples is horizontally distributed across several data sources. The algorithm requires only one pass through each remote source of training examples, and its accuracy and efficiency follow a clear pattern as function of a user-defined parameter. We outline an agent-based implementation of the algorithm.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Hofmann, T., Schölkopf, B., Smola, A.J.: Kernel methods in machine learning. Annals of Statistics 36(3), 1171–1220 (2008)MathSciNetCrossRefMATHGoogle Scholar
  2. 2.
    Caragea, D.: Learning Classifiers from Distributed, Semantically Heterogeneous, Autonomous Data Sources. PhD thesis, Iowa State University (2004)Google Scholar
  3. 3.
    Hazan, T., Man, A., Shashua, A.: A parallel decomposition solver for svm: Distributed dual ascend using fenchel duality, pp. 1–8 (June 2008)Google Scholar
  4. 4.
    Schölkopf, B., Smola, A.J.: Learning with Kernels: Support Vector Machines, Regularization, Optimization, and Beyond. MIT Press, Cambridge (2001)Google Scholar
  5. 5.
    Tsang, I., Kwok, J., Cheung, P.M.: Core vector machines: Fast svm training on very large data sets. Journal of Machine Learning Research 6, 363–392 (2005)MathSciNetMATHGoogle Scholar
  6. 6.
    Bădoiu, M., Clarkson, K.L.: Optimal core-sets for balls. Comput. Geom. Theory Appl. 40(1), 14–22 (2008)MathSciNetCrossRefMATHGoogle Scholar
  7. 7.
    Syed, N.A., Liu, H., Sung, K.K.: Incremental learning with support vector machines. In: Proc. of the 16th IJCAI (1999)Google Scholar
  8. 8.
    Asuncion, A., Newman, D.: UCI machine learning repository (2007), http://www.ics.uci.edu/~mlearn/MLRepository.html

Copyright information

© Springer-Verlag Berlin Heidelberg 2009

Authors and Affiliations

  • Stefano Lodi
    • 1
  • Ricardo Ñanculef
    • 2
  • Claudio Sartori
    • 1
  1. 1.Dept. of Electronics, Comp. Sc. and SystemsUniversity of BolognaItaly
  2. 2.Department of InformaticsFederico Santa María UniversityChile

Personalised recommendations