Tackling unsupervised multi-source domain adaptation with optimism and consistency

作者：

Highlights：

• Multi-source data is weighted dynamically through a theoretically-derived objective.

• Negative transfer is mitigated through consistency regularization on target data.

• Achieves state of the art results in multiple benchmark datasets.

摘要

•Multi-source data is weighted dynamically through a theoretically-derived objective.•Negative transfer is mitigated through consistency regularization on target data.•Achieves state of the art results in multiple benchmark datasets.

论文关键词：Multi-source domain adaptation,Transfer learning,Adversarial learning,Consistency regularization

论文评审过程：Received 8 December 2020, Revised 1 October 2021, Accepted 28 December 2021, Available online 15 January 2022, Version of Record 21 January 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.116486