SampleRank Training for Phrase-Based Machine Translation

Statistical machine translation systems are normally optimised for a chosen gain function (metric) by using MERT to find the best model weights. This algorithm suffers from stability problems and cannot scale beyond 20-30 features. We present an alternative algorithm for discriminative training of phrasebasedMT systems, SampleRank, which scales to hundreds of features, equals or beats MERT on both small and medium sized systems, and permits the use of sentence or document level features. SampleRank proceeds by repeatedly updating the model weights to ensure that the ranking of output sentences induced by the model is the same as that induced by the gain function.