Abstract
This paper highlights the impressive utility of multi-parallel corpora for transfer learning in a one-to-many low-resource neural machine translation (NMT) setting. We report on a systematic comparison of multistage fine-tuning configurations, consisting of (1) pre-training on an external large (209k–440k) parallel corpus for English and a helping target language, (2) mixed pre-training or fine-tuning on a mixture of the external and low-resource (18k) target parallel corpora, and (3) pure fine-tuning on the target parallel corpora. Our experiments confirm that multi-parallel corpora are extremely useful despite their scarcity and content-wise redundancy thus exhibiting the true power of multilingualism. Even when the helping target language is not one of the target languages of our concern, our multistage fine-tuning can give 3–9 BLEU score gains over a simple one-to-one model.- Anthology ID:
- D19-1146
- Volume:
- Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)
- Month:
- November
- Year:
- 2019
- Address:
- Hong Kong, China
- Venues:
- EMNLP | IJCNLP
- SIG:
- SIGDAT
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 1410–1416
- Language:
- URL:
- https://aclanthology.org/D19-1146
- DOI:
- 10.18653/v1/D19-1146
- Cite (ACL):
- Raj Dabre, Atsushi Fujita, and Chenhui Chu. 2019. Exploiting Multilingualism through Multistage Fine-Tuning for Low-Resource Neural Machine Translation. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 1410–1416, Hong Kong, China. Association for Computational Linguistics.
- Cite (Informal):
- Exploiting Multilingualism through Multistage Fine-Tuning for Low-Resource Neural Machine Translation (Dabre et al., EMNLP-IJCNLP 2019)
- PDF:
- https://preview.aclanthology.org/auto-file-uploads/D19-1146.pdf