Margin embedding net for robust margin collaborative representation-based classification

作者：

Highlights：

• In order to improve the performance of CRC, we extract more discriminant features and use more representative samples simultaneously.

• By analyzing the collaborative representation mechanism, we propose a simple but effective classification method based on robust marginal training samples, which is called Robust Margin Collaborative Representation based Classification method (RMCRC).

• A deep feature extraction method termed Margin Embedding Net (MEN) is proposed to enhance the performance of robust marginal training samples and has a close connection with RMCRC.

• A collaborative representation-based triplet mining mechanism for MEN is proposed.

摘要

•In order to improve the performance of CRC, we extract more discriminant features and use more representative samples simultaneously.•By analyzing the collaborative representation mechanism, we propose a simple but effective classification method based on robust marginal training samples, which is called Robust Margin Collaborative Representation based Classification method (RMCRC).•A deep feature extraction method termed Margin Embedding Net (MEN) is proposed to enhance the performance of robust marginal training samples and has a close connection with RMCRC.•A collaborative representation-based triplet mining mechanism for MEN is proposed.

论文关键词：Collaborative representation,Feature extraction,Marginal sample,Image classification

论文评审过程：Received 21 September 2021, Revised 26 July 2022, Accepted 20 August 2022, Available online 22 August 2022, Version of Record 31 August 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108991