KRNN: k Rare-class Nearest Neighbour classification. (February 2017)

• Record Type: Journal Article
• Authors: Zhang, Xiuzhen; Li, Yuxuan; Kotagiri, Ramamohanarao; Wu, Lifang; Tari, Zahir; Cheriet, Mohamed
• Is Part Of: Pattern recognition. Volume 62(2017:Feb.)
• Page Start: 33
• Page End: 44
• Publication Date: 2017-02
• Subjects: Imbalanced classification -- Nearest neighbour classification -- KNN -- Re-sampling -- Cost-sensitive learning; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2016.08.023
• Languages: English
• ISSNs: 0031-3203

Abstract

Abstract: Imbalanced classification is a challenging problem. Re-sampling and cost-sensitive learning are global strategies for generality-oriented algorithms such as the decision tree, targeting inter-class imbalance. We research local strategies for the specificity-oriented learning algorithms like the k Nearest Neighbour (KNN) to address the within-class imbalance issue of positive data sparsity. We propose an algorithm k Rare-class Nearest Neighbour, or KRNN, by directly adjusting the induction bias of KNN. We propose to form dynamic query neighbourhoods, and to further adjust the positive posterior probability estimation to bias classification towards the rare class. We conducted extensive experiments on thirty real-world and artificial datasets to evaluate the performance of KRNN. Our experiments showed that KRNN significantly improved KNN for classification of the rare class, and often outperformed re-sampling and cost-sensitive learning strategies with generality-oriented base learners. Abstract : Highlights: Nearest neighbour classification algorithm for accurate rare-class classification. Dynamic nearest neighbourhood formulation. Adjusted posterior class probability estimation biased for the rare class.