Artifacts of Markov blanket filtering based on discretized features in small sample size applications

Authors:
Theo A. Knijnenburg;Marcel J. T. Reinders;Lodewyk F. A. Wessels
Affiliations:
Information and Communication Theory Group, Faculty of Electrical Engineering, Mathematics and Computer Science, Delft University of Technology, Mekelweg 4, 2628 CD Delft, The Netherlands;Information and Communication Theory Group, Faculty of Electrical Engineering, Mathematics and Computer Science, Delft University of Technology, Mekelweg 4, 2628 CD Delft, The Netherlands;Information and Communication Theory Group, Faculty of Electrical Engineering, Mathematics and Computer Science, Delft University of Technology, Mekelweg 4, 2628 CD Delft, The Netherlands and Depa ...
Venue:
Pattern Recognition Letters
Year:
2006

Citing 2
Cited 1

Feature selection for high-dimensional genomic microarray data

ICML '01 Proceedings of the Eighteenth International Conference on Machine Learning
Iteratively Selecting Feature Subsets for Mining from High-Dimensional Databases

PKDD '02 Proceedings of the 6th European Conference on Principles of Data Mining and Knowledge Discovery

An extended Markov blanket approach to proteomic biomarker detection from high-resolution mass spectrometry data

IEEE Transactions on Information Technology in Biomedicine

Quantified Score

Hi-index	0.10

Visualization

Abstract

Markov blanket filtering based on discretized features (MBF) has been proposed as a feature selection strategy. Critical evaluation of MBF has demonstrated its contradictory and counterintuitive nature, which results in undesirable properties for small sample size applications such as classification based on microarray gene expression data.