Learning deep discriminative embeddings via joint rescaled features and log-probability centers

作者：

Highlights：

• Extend center loss to log probability space for better generalization.

• Modify center loss for better geometry interpretation.

• Rescale deep features for better discrimination.

• Theory analysis is provided for ground support.

摘要

•Extend center loss to log probability space for better generalization.•Modify center loss for better geometry interpretation.•Rescale deep features for better discrimination.•Theory analysis is provided for ground support.

论文关键词：Deep discriminative embedding,Softmax loss,Easing overfitting

论文评审过程：Received 19 November 2019, Revised 3 May 2020, Accepted 22 January 2021, Available online 27 January 2021, Version of Record 15 February 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2021.107852