VoteTRANS: Detecting Adversarial Text without Training by Voting on Hard Labels of Transformations
Hoang-Quoc Nguyen-Son, Seira Hidano, Kazuhide Fukushima, Shinsaku Kiyomoto, Isao Echizen
Abstract
Adversarial attacks reveal serious flaws in deep learning models. More dangerously, these attacks preserve the original meaning and escape human recognition. Existing methods for detecting these attacks need to be trained using original/adversarial data. In this paper, we propose detection without training by voting on hard labels from predictions of transformations, namely, VoteTRANS. Specifically, VoteTRANS detects adversarial text by comparing the hard labels of input text and its transformation. The evaluation demonstrates that VoteTRANS effectively detects adversarial text across various state-of-the-art attacks, models, and datasets.- Anthology ID:
- 2023.findings-acl.315
- Volume:
- Findings of the Association for Computational Linguistics: ACL 2023
- Month:
- July
- Year:
- 2023
- Address:
- Toronto, Canada
- Venue:
- Findings
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 5090–5104
- Language:
- URL:
- https://aclanthology.org/2023.findings-acl.315
- DOI:
- 10.18653/v1/2023.findings-acl.315
- Cite (ACL):
- Hoang-Quoc Nguyen-Son, Seira Hidano, Kazuhide Fukushima, Shinsaku Kiyomoto, and Isao Echizen. 2023. VoteTRANS: Detecting Adversarial Text without Training by Voting on Hard Labels of Transformations. In Findings of the Association for Computational Linguistics: ACL 2023, pages 5090–5104, Toronto, Canada. Association for Computational Linguistics.
- Cite (Informal):
- VoteTRANS: Detecting Adversarial Text without Training by Voting on Hard Labels of Transformations (Nguyen-Son et al., Findings 2023)
- PDF:
- https://preview.aclanthology.org/remove-xml-comments/2023.findings-acl.315.pdf