Feature Selection for High-Dimensional Data: A Kolmogorov-Smirnov Correlation-Based Filter.


Jacek Biesiada1, Wlodzislaw Duch2,3,
1Division of Computer Studies, Department of Electrotechnology, The Silesian University of Technology, Katowice, Poland;
2School of Computer Engineering, Nanyang Technological University, Singapore,
3Department of Informatics, Nicolaus Copernicus University, Grudziadzka 5, 87-100 Torun, Poland.

Abstract.

An algorithm for filtering information based on the Kolmogorov-Smirnov correlation- based approach has been implemented and tested on feature selection. The only parameter of this algorithm is statistical confidence level that two distributions are identical. Empirical comparisons with 4 other state-of-the-art features selection algorithms (FCBF, CorrSF, ReliefF and ConnSF) are very encouraging.

Reference: Biesiada J, Duch W (2005), Feature Selection for High-Dimensional Data: A Kolmogorov-Smirnov Correlation-Based Filter.
Advances in Soft Computing, Springer Verlag, pp. 95-104, 2005.
Also: Computer Recognition Systems. Proc. of the 4th International Conference on Computer Recognition Systems (CORES'05), Ed. M. Kurzynski, E. Puchała, M. Wozniak, A. Zolnierek,

Preprint for comments in PDF, 72 KB.

BACK to the publications of W. Duch.
BACK to the on-line publications of the Department of Informatics, NCU.