Feature selection for dimensionality reduction

Abstract

Dimensionality reduction is a commonly used step in machine learning, especially when dealing with a high dimensional space of features. The original feature space is mapped onto a new, reduced dimensionally space. The dimensionality reduction is usually performed either by selecting a subset of the original dimensions or/and by constructing new dimensions. This paper deals with feature subset selection for dimensionality reduction in machine learning. We provide a brief overview of the feature subset selection techniques that are commonly used in machine learning. Detailed description is provided for feature sub-set selection as commonly used on text data. For illustration, we show performance of several methods on document categorization of real-world data. © Springer-Verlag Berlin Heidelberg 2006.