Matching Distributions between Model and Data: Cross-domain Knowledge Distillation for Unsupervised Domain Adaptation

Bo Zhang, Xiaoming Zhang, Yun Liu, Lei Cheng, Zhoujun Li


Abstract
Unsupervised Domain Adaptation (UDA) aims to transfer the knowledge of source domain to the unlabeled target domain. Existing methods typically require to learn to adapt the target model by exploiting the source data and sharing the network architecture across domains. However, this pipeline makes the source data risky and is inflexible for deploying the target model. This paper tackles a novel setting where only a trained source model is available and different network architectures can be adapted for target domain in terms of deployment environments. We propose a generic framework named Cross-domain Knowledge Distillation (CdKD) without needing any source data. CdKD matches the joint distributions between a trained source model and a set of target data during distilling the knowledge from the source model to the target domain. As a type of important knowledge in the source domain, for the first time, the gradient information is exploited to boost the transfer performance. Experiments on cross-domain text classification demonstrate that CdKD achieves superior performance, which verifies the effectiveness in this novel setting.
Anthology ID:
2021.acl-long.421
Volume:
Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)
Month:
August
Year:
2021
Address:
Online
Editors:
Chengqing Zong, Fei Xia, Wenjie Li, Roberto Navigli
Venues:
ACL | IJCNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
5423–5433
Language:
URL:
https://aclanthology.org/2021.acl-long.421
DOI:
10.18653/v1/2021.acl-long.421
Bibkey:
Cite (ACL):
Bo Zhang, Xiaoming Zhang, Yun Liu, Lei Cheng, and Zhoujun Li. 2021. Matching Distributions between Model and Data: Cross-domain Knowledge Distillation for Unsupervised Domain Adaptation. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 5423–5433, Online. Association for Computational Linguistics.
Cite (Informal):
Matching Distributions between Model and Data: Cross-domain Knowledge Distillation for Unsupervised Domain Adaptation (Zhang et al., ACL-IJCNLP 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.acl-long.421.pdf
Video:
 https://aclanthology.org/2021.acl-long.421.mp4