Iterative Refinement for Machine Translation

Roman Novak, Michael Auli, David Grangier

Nov 02, 2016 (modified: Dec 15, 2016) ICLR 2017 conference submission readers: everyone
  • Abstract: Existing machine translation decoding algorithms generate translations in a strictly monotonic fashion and never revisit previous decisions. As a result, earlier mistakes cannot be corrected at a later stage. In this paper, we present a translation scheme that starts from an initial guess and then makes iterative improvements that may revisit previous decisions. We parameterize our model as a convolutional neural network that predicts discrete substitutions to an existing translation based on an attention mechanism over both the source sentence as well as the current translation output. By making less than one modification per sentence, we improve the output of a phrase-based translation system by up to 0.4 BLEU on WMT15 German-English translation.
  • TL;DR: We propose of novel decoding strategy for MT, after producing a full sentence the model can revisit its choice and substitute words; multiple words can iteratively be edited.
  • Keywords: Natural language processing
  • Conflicts:,,