Normalized class coherence change-based kNN for classification of imbalanced data
作者：
Highlights：
• NCC-kNN is a k nearest neighbor classification algorithm for imbalanced classification.
• NCC-kNN considers not only imbalance ratio but also the difference in class coherence by classes.
• NCC-kNN normalizes the average class coherence change by the mean class coherence of each class.
• NCC-kNN considers the class coherence of a new sample with the mean class coherence as the base.
• NCC-kNN is prone to detect the rare class by considering the low density in the rare class.
摘要
•NCC-kNN is a k nearest neighbor classification algorithm for imbalanced classification.•NCC-kNN considers not only imbalance ratio but also the difference in class coherence by classes.•NCC-kNN normalizes the average class coherence change by the mean class coherence of each class.•NCC-kNN considers the class coherence of a new sample with the mean class coherence as the base.•NCC-kNN is prone to detect the rare class by considering the low density in the rare class.
论文关键词：kNN,Nearest neighbor classification,Imbalanced data,Class coherence
论文评审过程：Received 24 May 2019, Revised 12 June 2021, Accepted 23 June 2021, Available online 29 June 2021, Version of Record 24 July 2021.
论文官网地址：https://doi.org/10.1016/j.patcog.2021.108126
