Enhancing Out-of-Distribution Detection in Natural Language Understanding via Implicit Layer Ensemble
Hyunsoo Cho, Choonghyun Park, Jaewook Kang, Kang Min Yoo, Taeuk Kim, Sang-goo Lee
Abstract
Out-of-distribution (OOD) detection aims to discern outliers from the intended data distribution, which is crucial to maintaining high reliability and a good user experience. Most recent studies in OOD detection utilize the information from a single representation that resides in the penultimate layer to determine whether the input is anomalous or not. Although such a method is straightforward, the potential of diverse information in the intermediate layers is overlooked. In this paper, we propose a novel framework based on contrastive learning that encourages intermediate features to learn layer-specialized representations and assembles them implicitly into a single representation to absorb rich information in the pre-trained language model. Extensive experiments in various intent classification and OOD datasets demonstrate that our approach is significantly more effective than other works.- Anthology ID:
- 2022.findings-emnlp.55
- Volume:
- Findings of the Association for Computational Linguistics: EMNLP 2022
- Month:
- December
- Year:
- 2022
- Address:
- Abu Dhabi, United Arab Emirates
- Editors:
- Yoav Goldberg, Zornitsa Kozareva, Yue Zhang
- Venue:
- Findings
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 783–798
- Language:
- URL:
- https://aclanthology.org/2022.findings-emnlp.55
- DOI:
- 10.18653/v1/2022.findings-emnlp.55
- Cite (ACL):
- Hyunsoo Cho, Choonghyun Park, Jaewook Kang, Kang Min Yoo, Taeuk Kim, and Sang-goo Lee. 2022. Enhancing Out-of-Distribution Detection in Natural Language Understanding via Implicit Layer Ensemble. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 783–798, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
- Cite (Informal):
- Enhancing Out-of-Distribution Detection in Natural Language Understanding via Implicit Layer Ensemble (Cho et al., Findings 2022)
- PDF:
- https://preview.aclanthology.org/ingest-acl-2023-videos/2022.findings-emnlp.55.pdf