Improving Neural Machine Translation Models with Monolingual Data

ACL 2016  ·  Rico Sennrich, Barry Haddow, Alexandra Birch ·

Neural Machine Translation (NMT) has obtained state-of-the art performance for several language pairs, while only using parallel data for training. Target-side monolingual data plays an important role in boosting fluency for phrase-based statistical machine translation, and we investigate the use of monolingual data for NMT. In contrast to previous work, which combines NMT models with separately trained language models, 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 train with monolingual data without changing the neural network architecture. By pairing monolingual training data with an automatic back-translation, we can treat it as additional parallel training data, and we obtain substantial improvements on the WMT 15 task English<->German (+2.8-3.7 BLEU), 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 English->German.

PDF Abstract ACL 2016 PDF ACL 2016 Abstract

Datasets


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Cross-Lingual Bitext Mining BUCC French-to-English Monolingual training data F1 score 75.8 # 3
Cross-Lingual Bitext Mining BUCC German-to-English Monolingual training data F1 score 76.9 # 3

Methods


No methods listed for this paper. Add relevant methods here