EPRENNID: An evolutionary prototype reduction based ensemble for nearest neighbor classification of imbalanced data

Vluymans, Sarah; Triguero, Isaac; Cornelis, Chris; Saeys, Yvan

Abstract

Classification problems with an imbalanced class distribution have received an increased amount of attention within the machine learning community over the last decade. They are encountered in a growing number of real-world situations and pose a challenge to standard machine learning techniques. We propose a new hybrid method specifically tailored to handle class imbalance, called EPRENNID. It performs an evolutionary prototype reduction focused on providing diverse solutions to prevent the method from overfitting the training set. It also allows us to explicitly reduce the underrepresented class, which the most common preprocessing solutions handling class imbalance usually protect. As part of the experimental study, we show that the proposed prototype reduction method outperforms state-of-the-art preprocessing techniques. The preprocessing step yields multiple prototype sets that are later used in an ensemble, performing a weighted voting scheme with the nearest neighbor classifier. EPRENNID is experimentally shown to significantly outperform previous proposals.

Citation

Vluymans, S., Triguero, I., Cornelis, C., & Saeys, Y. (2016). EPRENNID: An evolutionary prototype reduction based ensemble for nearest neighbor classification of imbalanced data. Neurocomputing, 216, https://doi.org/10.1016/j.neucom.2016.08.026

• Journal Article Type: Article
• Acceptance Date: Aug 4, 2016
• Online Publication Date: Aug 11, 2016
• Publication Date: Dec 5, 2016
• Deposit Date: Aug 26, 2016
• Publicly Available Date: Aug 26, 2016
• Journal: Neurocomputing
• Print ISSN: 0925-2312
• Electronic ISSN: 1872-8286
• Publisher: Elsevier
• Peer Reviewed: Peer Reviewed
• Volume: 216
• DOI: https://doi.org/10.1016/j.neucom.2016.08.026
• Keywords: Imbalanced data; Prototype selection; Prototype generation; ; Differential evolution; Nearest neighbor
• Public URL: https://nottingham-repository.worktribe.com/output/836565
• Publisher URL: http://www.sciencedirect.com/science/article/pii/S0925231216308669
• Contract Date: Aug 26, 2016