Abstract
Currently, adversarial training has become a popular and powerful regularization method in the natural language domain. In this paper, we Regularized Adversarial Training (R-AT) via dropout, which forces the output probability distributions of different sub-models generated by dropout to be consistent under the same adversarial samples. Specifically, we generate adversarial samples by perturbing the word embeddings. For each adversarial sample fed to the model, R-AT minimizes both the adversarial risk and the bidirectional KL-divergence between the adversarial output distributions of two sub-models sampled by dropout. Through extensive experiments on 13 public natural language understanding datasets, we found that R-AT has improvements for many models (e.g., rnn-based, cnn-based, and transformer-based models). For the GLUE benchmark, when R-AT is only applied to the fine-tuning stage, it is able to improve the overall test score of the BERT-base model from 78.3 to 79.6 and the RoBERTa-large model from 88.1 to 88.6. Theoretical analysis reveals that R-AT has potential gradient regularization during the training process. Furthermore, R-AT can reduce the inconsistency between training and testing of models with dropout.- Anthology ID:
- 2022.findings-emnlp.480
- 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:
- 6427–6440
- Language:
- URL:
- https://aclanthology.org/2022.findings-emnlp.480
- DOI:
- 10.18653/v1/2022.findings-emnlp.480
- Cite (ACL):
- Shiwen Ni, Jiawen Li, and Hung-Yu Kao. 2022. R-AT: Regularized Adversarial Training for Natural Language Understanding. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 6427–6440, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
- Cite (Informal):
- R-AT: Regularized Adversarial Training for Natural Language Understanding (Ni et al., Findings 2022)
- PDF:
- https://preview.aclanthology.org/dois-2013-emnlp/2022.findings-emnlp.480.pdf