# Adversarial Subword Regularization for Robust Neural Machine Translation

arXiv: https://arxiv.org/abs/2004.14109 · HTML: https://mlsanity.com/paper/2004.14109/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 29 Apr 2020.

## Built on by (most cited first, 6 of 6)

- [TransFool: An Adversarial Attack against Neural Machine Translation Models](https://mlsanity.com/paper/2302.00944/) (2023)
- [Consistency Training with Virtual Adversarial Discrete Perturbation](https://mlsanity.com/paper/2104.07284/) (2021)
- [Did Translation Models Get More Robust Without Anyone Even Noticing?](https://mlsanity.com/paper/2403.03923/) (2024)
- [Synthetic Data Generation and Joint Learning for Robust Code-Mixed Translation](https://mlsanity.com/paper/2403.16771/) (2024)
- [Misspellings in Natural Language Processing: A survey](https://mlsanity.com/paper/2501.16836/) (2025)
- [Pre-Editorial Normalization for Automatically Transcribed Medieval Manuscripts in Old French and Latin](https://mlsanity.com/paper/2602.13905/) (2026)

## Abstract (arXiv)

Exposing diverse subword segmentations to neural machine translation (NMT) models often improves the robustness of machine translation as NMT models can experience various subword candidates. However, the diversification of subword segmentations mostly relies on the pre-trained subword language models from which erroneous segmentations of unseen words are less likely to be sampled. In this paper, we present adversarial subword regularization (ADVSR) to study whether gradient signals during training can be a substitute criterion for exposing diverse subword segmentations. We experimentally show that our model-based adversarial samples effectively encourage NMT models to be less sensitive to segmentation errors and improve the performance of NMT models in low-resource and out-domain datasets.