Projects per year
Abstract
Neural Machine Translation (NMT) has obtained state-of-the art performance for several language pairs, while only using parallel data for training. Monolingual data plays an important role in boosting fluency for phrase-based statistical machine translation, and we investigate the use of monolingual data for neural machine translation (NMT). In contrast to previous work, which integrates a separately trained RNN language model into an NMT architecture (Gülçehre et al., 2015), we note that encoder-decoder NMT architectures already have the capacity to learn the same information as a language model, and we explore strategies to include monolingual training data in the training process. Through our use of monolingual data, we obtain substantial improvements on the WMT 15 (+2.8–3.4 BLEU) task for English↔German, and for the low-resourced IWSLT 14 task Turkish!English (+2.1–3.4 BLEU), obtaining new state-of-the-art results. We also show that fine-tuning on in-domain monolingual and parallel data gives substantial improvements for the IWSLT 15 task for English→German.
Original language | English |
---|---|
Title of host publication | Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics |
Place of Publication | Berlin, Germany |
Publisher | Association for Computational Linguistics (ACL) |
Pages | 86-96 |
Number of pages | 11 |
Volume | 1: Long Papers |
ISBN (Print) | 978-1-945626-00-5 |
DOIs | |
Publication status | Published - 12 Aug 2016 |
Event | 54th Annual Meeting of the Association for Computational Linguistics - Berlin, Germany Duration: 7 Aug 2016 → 12 Aug 2016 https://mirror.aclweb.org/acl2016/ |
Conference
Conference | 54th Annual Meeting of the Association for Computational Linguistics |
---|---|
Abbreviated title | ACL 2016 |
Country/Territory | Germany |
City | Berlin |
Period | 7/08/16 → 12/08/16 |
Internet address |
Fingerprint
Dive into the research topics of 'Improving Neural Machine Translation Models with Monolingual Data'. Together they form a unique fingerprint.Projects
- 1 Finished