An improvement direction for filter selection techniques using information theory measures and quadratic optimization

TL;DR

This paper proposes a mathematical optimization approach using information theory and quadratic optimization to reduce feature redundancy and improve the effectiveness of filter selection techniques in classification tasks.

Contribution

It introduces a novel optimization-based method that enhances filter selection by explicitly minimizing feature redundancy while maximizing relevance.

Findings

Reduces feature inter-redundancy effectively

Improves classification model generalization

Outperforms traditional filter methods

Abstract

Filter selection techniques are known for their simplicity and efficiency. However this kind of methods doesn't take into consideration the features inter-redundancy. Consequently the un-removed redundant features remain in the final classification model, giving lower generalization performance. In this paper we propose to use a mathematical optimization method that reduces inter-features redundancy and maximize relevance between each feature and the target variable.