The MIT-LL/AFRL IWSLT-2013 MT System


This paper describes the MIT-LL/AFRL statistical MT system and the improvements that were developed during the IWSLT 2013 evaluation campaign [1]. As part of these efforts, we experimented with a number of extensions to the standard phrase-based model that improve performance on the Russian to English, Chinese to English, Arabic to English, and English to French TED-talk translation task. We also applied our existing ASR system to the TED-talk lecture ASR task. We discuss the architecture of the MIT-LL/AFRL MT system, improvements over our 2012 system, and experiments we ran during the IWSLT-2013 evaluation. Specifically, we focus on 1) crossentropy filtering of MT training data, and 2) improved optimization techniques, 3) language modeling, and 4) approximation of out-ofvocabulary words.

Extracted Key Phrases

18 Figures and Tables

Cite this paper

@inproceedings{Kazi2013TheMI, title={The MIT-LL/AFRL IWSLT-2013 MT System}, author={Michaeel Kazi and Michael Coury and Elizabeth Salesky and Jessica Ray and Wade Shen and Terry Gleason and Tim Anderson and Grant Erdmann and Lane Schwartz and Brian Ore and Raymond Slyh and Jeremy Gwinnup and Katherine Young and Michael Hutt}, year={2013} }