Discovering and incorporating latent target-domains for domain adaptation

作者：

Highlights：

• We focus on the practical problem in unsupervised domain adaptation that multiple latent domains are observed in the target domain.

• We propose a domain discovery scheme based on the characteristics of the target domain and the relationship between source and target domains.

• We propose a method to jointly learn the mapping function based on multiple latent domains.

摘要

•We focus on the practical problem in unsupervised domain adaptation that multiple latent domains are observed in the target domain.•We propose a domain discovery scheme based on the characteristics of the target domain and the relationship between source and target domains.•We propose a method to jointly learn the mapping function based on multiple latent domains.

论文关键词：Unsupervised domain adaptation,Latent domain,Multiple kernel learning

论文评审过程：Received 20 May 2019, Revised 30 June 2020, Accepted 1 July 2020, Available online 2 July 2020, Version of Record 26 July 2020.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107536