Neural Machine Translation with Pivot Languages
arXiv: Computation and Language, Volume abs/1611.04928, 2016.
Neural machine translation systems typically rely on the size of parallel corpora. Nevertheless, high-quality parallel corpora are scarce resources for specific language pairs and domains. For a source-to-target language pair with a small parallel corpus, we introduce the pivot language to bridge source language and target language under ...More
Full Text (Upload PDF)
PPT (Upload PPT)