Computer Sciences


 Feature Selection in Pattern Recognition

This page is obsolete! Please go to
the new department website.

Comparison of statistical and evolutionary approaches to feature selection in pattern recognition (Aktion Project 23p20)

General Information

Contact person: Helmut A. Mayer


This Czech-Austrian joint research project aims to combine the knowledge of the Czech partners in statistical feature selection with the experience of the Austrian partners in evolutionary computation. In this start project the various algorithms are to be compared based on statistical measures of the generated feature subsets, and on classification error of commonly used classifiers (e.g., Gaussian, k-NN, ANN). Especially, the potential of these methods to reduce the dimensionality of large scale problems are to be studied.

More information about this project is available at

Last modified: 2013-09-02 14:07. Webmaster contact: