Cluster-based oversampling with area extraction from representative points for class imbalance learning

Farou, Zakarya; Wang, Yizhi; Horváth, Tomáš

Abstract

Class imbalance learning is challenging in various domains where training datasets exhibit disproportionate samples in a specific class. Resampling methods have been used to adjust the class distribution, but they often have limitations for small disjunct minority subsets. This paper introduces AROSS, an adaptive cluster-based oversampling approach that addresses these limitations. AROSS utilizes an optimized agglomerative clustering algorithm with the Cophenetic Correlation Coefficient and the Bayesian Information Criterion to identify representative areas of the minority class. Safe and half-safe areas are obtained using an incremental k-Nearest Neighbor strategy, and oversampling is performed with a truncated hyperspherical Gaussian distribution. Experimental evaluations on 70 binary datasets demonstrate the effectiveness of AROSS in improving class imbalance learning performance, making it a promising solution for mitigating class imbalance challenges, especially for small disjunct minority subsets.

Citation

Farou, Z., Wang, Y., & Horváth, T. (2024). Cluster-based oversampling with area extraction from representative points for class imbalance learning. Intelligent Systems with Applications, 22, Article 200357. https://doi.org/10.1016/j.iswa.2024.200357

• Journal Article Type: Article
• Acceptance Date: Mar 8, 2024
• Online Publication Date: Mar 16, 2024
• Publication Date: 2024-06
• Deposit Date: Mar 25, 2024
• Publicly Available Date: Mar 25, 2024
• Journal: Intelligent Systems with Applications
• Publisher: Elsevier
• Peer Reviewed: Peer Reviewed
• Volume: 22
• Article Number: 200357
• DOI: https://doi.org/10.1016/j.iswa.2024.200357
• Keywords: Artificial Intelligence; Computer Science Applications; Computer Vision and Pattern Recognition; Signal Processing; Computer Science (miscellaneous)
• Public URL: http://researchrepository.napier.ac.uk/Output/3575166