Doubly-Trained Adversarial Data Augmentation for Neural Machine Translation

Weiting Tan, Shuoyang Ding, Huda Khayrallah, Philipp Koehn


Abstract
Neural Machine Translation (NMT) models are known to suffer from noisy inputs. To make models robust, we generate adversarial augmentation samples that attack the model and preserve the source-side meaning at the same time. To generate such samples, we propose a doubly-trained architecture that pairs two NMT models of opposite translation directions with a joint loss function, which combines the target-side attack and the source-side semantic similarity constraint. The results from our experiments across three different language pairs and two evaluation metrics show that these adversarial samples improve model robustness.
Anthology ID:
2022.amta-research.12
Volume:
Proceedings of the 15th biennial conference of the Association for Machine Translation in the Americas (Volume 1: Research Track)
Month:
September
Year:
2022
Address:
Orlando, USA
Editors:
Kevin Duh, Francisco Guzmán
Venue:
AMTA
SIG:
Publisher:
Association for Machine Translation in the Americas
Note:
Pages:
157–174
Language:
URL:
https://aclanthology.org/2022.amta-research.12
DOI:
Bibkey:
Cite (ACL):
Weiting Tan, Shuoyang Ding, Huda Khayrallah, and Philipp Koehn. 2022. Doubly-Trained Adversarial Data Augmentation for Neural Machine Translation. In Proceedings of the 15th biennial conference of the Association for Machine Translation in the Americas (Volume 1: Research Track), pages 157–174, Orlando, USA. Association for Machine Translation in the Americas.
Cite (Informal):
Doubly-Trained Adversarial Data Augmentation for Neural Machine Translation (Tan et al., AMTA 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.amta-research.12.pdf
Code
 steventan0110/NMTModelAttack
Data
WMT 2014