Decoupled self-supervised label augmentation for fully-supervised image classification
作者:
Highlights:
• We decouple the image classification task into three tasks with different objectives.
• We introduce a nonlinear projection head to address the semantic variation.
• We introduce a gating mechanism for self-supervised feature representation learning.
摘要
•We decouple the image classification task into three tasks with different objectives.•We introduce a nonlinear projection head to address the semantic variation.•We introduce a gating mechanism for self-supervised feature representation learning.
论文关键词:Self-supervised label augmentation,Image classification,Multi-task learning
论文评审过程:Received 2 April 2021, Revised 22 September 2021, Accepted 13 October 2021, Available online 26 October 2021, Version of Record 5 November 2021.
论文官网地址:https://doi.org/10.1016/j.knosys.2021.107605