Adversarial Subword Regularization for Robust Neural Machine Translation
arXiv:2004.14109
Abstract
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.
arXiv comment9 pages,1 figure, Findings of EMNLP 2020
References in corpus (7)
- Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
- Convolutional Sequence to Sequence Learning
- Adaptive Computation Time for Recurrent Neural Networks
- Towards Crafting Text Adversarial Samples
- Domain Robustness in Neural Machine Translation
- Multilingual Neural Machine Translation With Soft Decoupled Encoding
- Learning to Segment Inputs for NMT Favors Character-Level Processing