Electrostatic Field Classifier for Deficient Data

  • Marcin Budka
  • Bogdan Gabrys
Part of the Advances in Intelligent and Soft Computing book series (AINSC, volume 57)


This paper investigates the suitability of recently developed models based on the physical field phenomena for classification of incomplete datasets. An original approach to exploiting incomplete training data with missing features and labels, involving extensive use of electrostatic charge analogy has been proposed. Classification of incomplete patterns has been investigated using a local dimensionality reduction technique, which aims at exploiting all available information rather than trying to estimate the missing values. The performance of all proposed methods has been tested on a number of benchmark datasets for a wide range of missing data scenarios and compared to the performance of some standard techniques.


Charge Redistribution Miss Data Problem Incomplete Dataset Reduce Feature Space Incomplete Pattern 
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.
    Asuncion, A., Newman, D.J.: UCI machine learning repository (2007)Google Scholar
  2. 2.
    Dempster, A.P., Laird, N.M., Rubin, D.B.: Maximum Likelihood from Incomplete Data via EM Algorithm. Journal of Royal Statistical Society 39(1), 1–38 (1977)Google Scholar
  3. 3.
    Gabrys, B.: Neuro-fuzzy approach to processing inputs with missing values in pattern recognition problems. International Journal of Approximate Reasoning 30(3), 149–179 (2002)Google Scholar
  4. 4.
    Hochreiter, S., Mozer, M.C., Obermayer, K.: Coulomb classifiers: Generalizing support vector machines via an analogy to electrostatic systems. In: Advances in Neural Information Processing Systems, vol. 15, pp. 545–552 (2003)Google Scholar
  5. 5.
    Outhwaite, W., Turner, S.P.: Handbook of Social Science Methodology. SAGE Publications Ltd., Thousand Oaks (2007)Google Scholar
  6. 6.
    Principe, J.C., Xu, D., Fisher, J.: Information theoretic learning. Unsupervised Adaptive Filtering, 265–319 (2000)Google Scholar
  7. 7.
    Rubin, D.B.: Inference and missing data. Biometrika 63(3), 581–592Google Scholar
  8. 8.
    Ruta, D., Gabrys, B.: Physical field models for pattern classification. Soft Computing 8(2), 126–141 (2003)Google Scholar
  9. 9.
    Ruta, D., Gabrys, B.: A Framework for Machine Learning based on Dynamic Physical Fields. Natural Computing Journal (2007)Google Scholar
  10. 10.
    Schafer, J.L., Graham, J.W.: Missing data: Our view of the state of the art. Psychological Methods 7(2), 147–177 (2002)Google Scholar
  11. 11.
    Torkkola, K.: Feature extraction by non parametric mutual information maximization. The Journal of Machine Learning Research 3, 1415–1438 (2003)Google Scholar
  12. 12.
    Tresp, V., Ahmad, S., Neuneier, R.: Training neural networks with deficient data. In: Advances in Neural Information Processing Systems, vol. 6, pp. 128–135 (1994)Google Scholar
  13. 13.
    Zurek, W.H.: Complexity, Entropy and Physics of Information. Westview (1989)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2009

Authors and Affiliations

  • Marcin Budka
    • 1
  • Bogdan Gabrys
    • 1
  1. 1.School of Design, Engineering & ComputingComputational Intelligence Research Group, Bournemouth UniversityUnited Kingdom

Personalised recommendations