Ant Colony Optimization based Support Vector Machine for Improved Classification of Unbalanced Datasets

Abstract

Classifying unbalanced datasets is a challenging task, and traditional Support Vector Machine (SVM) algorithms often struggle to achieve accurate results. In this paper, we propose an Ant Colony Optimization based Support Vector Machine (SVM-ACO) algorithm that outperforms SVM on unbalanced datasets. Traditional SVM algorithms tend to be biased towards the majority class, leading to suboptimal performance for the minority class. However, SVM-ACO addresses this issue by incorporating the ant colony optimization approach, which assigns different weights to samples based on their class membership, giving higher importance to the minority class. This enables SVM-ACO to achieve a more balanced classification performance and improved accuracy. We compare the accuracy achieved by SVM-ACO and SVM on various imbalanced datasets from the literature. The experimental results support the effectiveness of SVM-ACO in addressing the challenges of unbalanced datasets. By incorporating ant colony optimization, SVM-ACO achieves improved accuracy and provides a promising alternative for handling unbalanced classification problems.