Stanford Neural Machine Translation Systems for Spoken Language Domains
Neural Machine Translation (NMT), though recently developed, has shown promising results for various language pairs. Despite that, NMT has only been applied to mostly formal texts such as those in the WMT shared tasks. This work further explores the effectiveness of NMT in spoken language domains by participating in the MT track of the IWSLT 2015. We consider two scenarios: (a) how to adapt existing NMT systems to a new domain and (b) the generalization of NMT to low-resource language pairs. Our results demonstrate that using an existing NMT framework1, we can achieve competitive results in the aforementioned scenarios when translating from English to German and Vietnamese. Notably, we have advanced state-of-the-art results in the IWSLT EnglishGerman MT track by up to 5.2 BLEU points.
PDF AbstractDatasets
Results from the Paper
Task | Dataset | Model | Metric Name | Metric Value | Global Rank | Benchmark |
---|---|---|---|---|---|---|
Machine Translation | IWSLT2015 English-Vietnamese | LSTM+Attention+Ensemble | BLEU | 26.4 | # 10 |