Relearning ensemble selection based on new generated features

TL;DR

This paper introduces a novel ensemble selection framework that relearns base classifiers and generates new features, improving classification performance by leveraging a new feature generation process within ensemble methods.

Contribution

It proposes a new ensemble selection approach with relearning of classifiers and feature generation, enhancing existing ensemble techniques.

Findings

Outperforms state-of-the-art ensemble methods on benchmark datasets.

Uses multiple performance measures to validate improvements.

Demonstrates effectiveness of new feature generation in ensemble learning.

Abstract

The ensemble methods are meta-algorithms that combine several base machine learning techniques to increase the effectiveness of the classification. Many existing committees of classifiers use the classifier selection process to determine the optimal set of base classifiers. In this article, we propose the classifiers selection framework with relearning base classifiers. Additionally, we use in the proposed framework the new generated feature, which can be obtained after the relearning process. The proposed technique was compared with state-of-the-art ensemble methods using three benchmark datasets and one synthetic dataset. Four classification performance measures are used to evaluate the proposed method.