Monte Carlo techniques for phrase-based translation

Abhishek Arun, Barry Haddow, Philipp Koehn, Adam Lopez, Chris Dyer, Phil Blunsom

Research output: Contribution to journalArticlepeer-review

Abstract / Description of output

Recent advances in statistical machine translation have used approximate beam search for NP-complete inference within probabilistic translation models. We present an alternative approach of sampling from the posterior distribution defined by a translation model. We define a novel Gibbs sampler for sampling translations given a source sentence and show that it effectively explores this posterior distribution. In doing so we overcome the limitations of heuristic beam search and obtain theoretically sound solutions to inference problems such as finding the maximum probability translation and minimum risk training and decoding.
Original languageEnglish
Pages (from-to)103-121
Number of pages19
JournalMachine Translation
Issue number2
Publication statusPublished - 2010


Dive into the research topics of 'Monte Carlo techniques for phrase-based translation'. Together they form a unique fingerprint.

Cite this