University of rochester WMT 2017 NMT system submission

Abstract

We describe the neural machine translation system submitted by the University of Rochester to the Chinese-English language pair for the WMT 2017 news translation task. We applied unsupervised word and subword segmentation techniques and deep learning in order to address (i) the word segmentation problem caused by the lack of delimiters between words and phrases in Chinese and (ii) the morphological and syntactic differences between Chinese and English. We integrated promising recent developments in NMT, including back-translations, language model reranking, subword splitting and minimum risk tuning.