Unsupervised Out-of-Domain Detection via Pre-trained Transformers
Keyang Xu, Tongzheng Ren, Shikun Zhang, Yihao Feng, Caiming Xiong
Abstract
Deployed real-world machine learning applications are often subject to uncontrolled and even potentially malicious inputs. Those out-of-domain inputs can lead to unpredictable outputs and sometimes catastrophic safety issues. Prior studies on out-of-domain detection require in-domain task labels and are limited to supervised classification scenarios. Our work tackles the problem of detecting out-of-domain samples with only unsupervised in-domain data. We utilize the latent representations of pre-trained transformers and propose a simple yet effective method to transform features across all layers to construct out-of-domain detectors efficiently. Two domain-specific fine-tuning approaches are further proposed to boost detection accuracy. Our empirical evaluations of related methods on two datasets validate that our method greatly improves out-of-domain detection ability in a more general scenario.- Anthology ID:
- 2021.acl-long.85
- 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
- Venues:
- ACL | IJCNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 1052–1061
- Language:
- URL:
- https://aclanthology.org/2021.acl-long.85
- DOI:
- 10.18653/v1/2021.acl-long.85
- Cite (ACL):
- Keyang Xu, Tongzheng Ren, Shikun Zhang, Yihao Feng, and Caiming Xiong. 2021. Unsupervised Out-of-Domain Detection via Pre-trained Transformers. 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 1052–1061, Online. Association for Computational Linguistics.
- Cite (Informal):
- Unsupervised Out-of-Domain Detection via Pre-trained Transformers (Xu et al., ACL-IJCNLP 2021)
- PDF:
- https://preview.aclanthology.org/ingestion-script-update/2021.acl-long.85.pdf
- Code
- rivercold/BERT-unsupervised-OOD
- Data
- SNLI, SST