Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidates

Taku Kudo


Abstract
Subword units are an effective way to alleviate the open vocabulary problems in neural machine translation (NMT). While sentences are usually converted into unique subword sequences, subword segmentation is potentially ambiguous and multiple segmentations are possible even with the same vocabulary. The question addressed in this paper is whether it is possible to harness the segmentation ambiguity as a noise to improve the robustness of NMT. We present a simple regularization method, subword regularization, which trains the model with multiple subword segmentations probabilistically sampled during training. In addition, for better subword sampling, we propose a new subword segmentation algorithm based on a unigram language model. We experiment with multiple corpora and report consistent improvements especially on low resource and out-of-domain settings.
Anthology ID:
P18-1007
Volume:
Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
Month:
July
Year:
2018
Address:
Melbourne, Australia
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
66–75
Language:
URL:
https://aclanthology.org/P18-1007
DOI:
10.18653/v1/P18-1007
Bibkey:
Cite (ACL):
Taku Kudo. 2018. Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidates. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 66–75, Melbourne, Australia. Association for Computational Linguistics.
Cite (Informal):
Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidates (Kudo, ACL 2018)
Copy Citation:
PDF:
https://aclanthology.org/P18-1007.pdf
Video:
 https://vimeo.com/285807834
Code
 google/sentencepiece +  additional community code