Smooth Soft-Balance Discriminative Analysis for imbalanced data
作者：
Highlights：
• S2BDA is for a difficult pattern recognition task: imbalance classification.
• Obtains clear boundaries between classes via smoothing techniques.
• Proposes soft-balance clustering to mine structures of data with theoretical proof.
• Proposes subclass-aware discriminant analysis to extract discriminative features.
• Various data sets (various imbalance ratio, data size and dimensionality) are tested.
摘要
•S2BDA is for a difficult pattern recognition task: imbalance classification.•Obtains clear boundaries between classes via smoothing techniques.•Proposes soft-balance clustering to mine structures of data with theoretical proof.•Proposes subclass-aware discriminant analysis to extract discriminative features.•Various data sets (various imbalance ratio, data size and dimensionality) are tested.
论文关键词：Imbalance classification,Smoothing,Soft-balanced clustering,Discriminative analysis
论文评审过程：Received 15 July 2020, Revised 2 November 2020, Accepted 6 November 2020, Available online 22 March 2021, Version of Record 7 July 2021.
论文官网地址：https://doi.org/10.1016/j.knosys.2020.106604
