Doubly-Trained Adversarial Data Augmentation for Neural Machine Translation
arXiv:2110.05691
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 semantic 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 the model robustness.
References in corpus (7)
- fairseq: A Fast, Extensible Toolkit for Sequence Modeling
- Robust Neural Machine Translation with Doubly Adversarial Inputs
- On Evaluation of Adversarial Perturbations for Sequence-to-Sequence Models
- Findings of the First Shared Task on Machine Translation Robustness
- AdvAug: Robust Adversarial Augmentation for Neural Machine Translation
- Reward Optimization for Neural Machine Translation with Learned Metrics
- Rethinking Perturbations in Encoder-Decoders for Fast Training