Scalable Feature Selection for Multi-class Problems
Boris Chidlovskii and Loic Lecerf
In: ECML/PKDD 2008, 15019 September 2008, Antwerp, Belgium.

Abstract

Scalable feature selection algorithms should remove irrelevant and redundant features and scale well on very large datasets. We identify that the currently best state-of-art methods perform well on binary classifica tion tasks but often underperform on multi-class tasks. We suggest that they suffer from the so-called accum ulative effect which becomes more visible with the growing number of classes and results in removing relevan t and unredundant features. To remedy the problem, we propose two new feature filtering methods which are bo th scalable and well adapted for the multi-class cases. We report the evaluation results on 17 different dat asets which include both binary and multi-class cases.

[Edit]