The UMD Neural Machine Translation Systems at WMT17 Bandit Learning Task
This addresses domain adaptation with limited feedback for machine translation practitioners, but is incremental as it builds on standard methods.
The paper tackled adapting a neural machine translation system to a new domain using only bandit feedback, achieving competitive results in the WMT17 German-English task by combining reinforcement learning and data selection.
We describe the University of Maryland machine translation systems submitted to the WMT17 German-English Bandit Learning Task. The task is to adapt a translation system to a new domain, using only bandit feedback: the system receives a German sentence to translate, produces an English sentence, and only gets a scalar score as feedback. Targeting these two challenges (adaptation and bandit learning), we built a standard neural machine translation system and extended it in two ways: (1) robust reinforcement learning techniques to learn effectively from the bandit feedback, and (2) domain adaptation using data selection from a large corpus of parallel data.