Abstract
Before entering the neural network, a token needs to be converted to its one-hot representation, which is a discrete distribution of the vocabulary. Smoothed representation is the probability of candidate tokens obtained from the pre-trained masked language model, which can be seen as a more informative augmented substitution to the one-hot representation. We propose an efficient data augmentation method, dub as text smoothing, by converting a sentence from its one-hot representation to controllable smoothed representation.We evaluate text smoothing on different datasets in a low-resource regime. Experimental results show that text smoothing outperforms various mainstream data augmentation methods by a substantial margin. Moreover, text smoothing can be combined with these data augmentation methods to achieve better performance.- Anthology ID:
- 2022.acl-short.97
- Volume:
- Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)
- Month:
- May
- Year:
- 2022
- Address:
- Dublin, Ireland
- Venue:
- ACL
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 871–875
- Language:
- URL:
- https://aclanthology.org/2022.acl-short.97
- DOI:
- 10.18653/v1/2022.acl-short.97
- Cite (ACL):
- Xing Wu, Chaochen Gao, Meng Lin, Liangjun Zang, and Songlin Hu. 2022. Text Smoothing: Enhance Various Data Augmentation Methods on Text Classification Tasks. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 871–875, Dublin, Ireland. Association for Computational Linguistics.
- Cite (Informal):
- Text Smoothing: Enhance Various Data Augmentation Methods on Text Classification Tasks (Wu et al., ACL 2022)
- PDF:
- https://preview.aclanthology.org/ingestion-script-update/2022.acl-short.97.pdf
- Code
- caskcsg/TextSmoothing
- Data
- SNIPS, SST