# Metalearning for Feature Selection

## TL;DR

This paper introduces a metalearning approach for feature selection that estimates feature quality based on related problems, significantly improving speed in supervised text classification tasks.

## Contribution

It proposes a novel framework for feature metalearning that integrates feature quality estimation into the selection process, enhancing efficiency.

## Key findings

- Feature metalearning can dramatically speed up feature selection.
- The approach outperforms standard heuristics in text classification.
- Extensive testing confirms the effectiveness of the method.

## Abstract

A general formulation of optimization problems in which various candidate solutions may use different feature-sets is presented, encompassing supervised classification, automated program learning and other cases. A novel characterization of the concept of a "good quality feature" for such an optimization problem is provided; and a proposal regarding the integration of quality based feature selection into metalearning is suggested, wherein the quality of a feature for a problem is estimated using knowledge about related features in the context of related problems. Results are presented regarding extensive testing of this "feature metalearning" approach on supervised text classification problems; it is demonstrated that, in this context, feature metalearning can provide significant and sometimes dramatic speedup over standard feature selection heuristics.