Pairwise Supervised Contrastive Learning of Sentence Representations

Dejiao Zhang, Shang-Wen Li, Wei Xiao, Henghui Zhu, Ramesh Nallapati, Andrew O. Arnold, Bing Xiang


Abstract
Many recent successes in sentence representation learning have been achieved by simply fine-tuning on the Natural Language Inference (NLI) datasets with triplet loss or siamese loss. Nevertheless, they share a common weakness: sentences in a contradiction pair are not necessarily from different semantic categories. Therefore, optimizing the semantic entailment and contradiction reasoning objective alone is inadequate to capture the high-level semantic structure. The drawback is compounded by the fact that the vanilla siamese or triplet losses only learn from individual sentence pairs or triplets, which often suffer from bad local optima. In this paper, we propose PairSupCon, an instance discrimination based approach aiming to bridge semantic entailment and contradiction understanding with high-level categorical concept encoding. We evaluate PairSupCon on various downstream tasks that involve understanding sentence semantics at different granularities. We outperform the previous state-of-the-art method with 10%–13% averaged improvement on eight clustering tasks, and 5%–6% averaged improvement on seven semantic textual similarity (STS) tasks.
Anthology ID:
2021.emnlp-main.467
Volume:
Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing
Month:
November
Year:
2021
Address:
Online and Punta Cana, Dominican Republic
Editors:
Marie-Francine Moens, Xuanjing Huang, Lucia Specia, Scott Wen-tau Yih
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
5786–5798
Language:
URL:
https://aclanthology.org/2021.emnlp-main.467
DOI:
10.18653/v1/2021.emnlp-main.467
Bibkey:
Cite (ACL):
Dejiao Zhang, Shang-Wen Li, Wei Xiao, Henghui Zhu, Ramesh Nallapati, Andrew O. Arnold, and Bing Xiang. 2021. Pairwise Supervised Contrastive Learning of Sentence Representations. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 5786–5798, Online and Punta Cana, Dominican Republic. Association for Computational Linguistics.
Cite (Informal):
Pairwise Supervised Contrastive Learning of Sentence Representations (Zhang et al., EMNLP 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.emnlp-main.467.pdf
Video:
 https://aclanthology.org/2021.emnlp-main.467.mp4
Code
 amazon-research/sentence-representations
Data
GLUEMPQA Opinion CorpusSICKSSTSentEval