Learning domain-shared group-sparse representation for unsupervised domain adaptation

作者：

Highlights：

• An equivalent condition for conditional distribution alignment is derived.

• Propose a domain-shared group-sparse dictionary learning model for unsupervised domain adaptation.

• Develop a classifier with both domain-shared and target-specific information.

• Our method achieves good performance in several cross-dataset recognition tasks.

摘要

•An equivalent condition for conditional distribution alignment is derived.•Propose a domain-shared group-sparse dictionary learning model for unsupervised domain adaptation.•Develop a classifier with both domain-shared and target-specific information.•Our method achieves good performance in several cross-dataset recognition tasks.

论文关键词：Domain adaptation,Dictionary learning

论文评审过程：Received 31 October 2017, Revised 27 February 2018, Accepted 27 April 2018, Available online 30 April 2018, Version of Record 24 May 2018.

论文官网地址：https://doi.org/10.1016/j.patcog.2018.04.027