Geometric imbalanced deep learning with feature scaling and boundary sample mining

作者：

Highlights：

• This paper considers the geometric distributions of samples and scale the features by the radius of hypersphere.

• This paper explores the relationships between samples, and then constrains the inter-class and the intra-class distances.

• The proposed model is evaluated on the three benchmark datasets and achieves the superior performance over previous methods.

摘要

•This paper considers the geometric distributions of samples and scale the features by the radius of hypersphere.•This paper explores the relationships between samples, and then constrains the inter-class and the intra-class distances.•The proposed model is evaluated on the three benchmark datasets and achieves the superior performance over previous methods.

论文关键词：Imbalance problem,Image classification,Geometric information,Boundary samples mining,Feature scaling

论文评审过程：Received 7 November 2020, Revised 1 December 2021, Accepted 29 January 2022, Available online 3 February 2022, Version of Record 6 February 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108564