We present an efficient framework to estimate the rule probabilities for a hierarchical phrasebased statistical machine translation system from parallel data. In previous work, this was done with bilingual parsing. We use a more efficient approach splitting the bilingual parsing into two stages, which allows us to train a hierarchical translation model on larger tasks. Furthermore, we apply lea...

The most popular algorithms for the estimation of the probabilities of a context-free grammar are the Inside-Outside algorithm and the Viterbi algorithm, which are Maximum Likelihood approaches. The diierence between the logarithm of the likelihood of a string and the logarithm of the likelihood of the most probable parse of a string is upper bounded linearly by the length of the string and the...

We present a phrasal inversion transduction grammar as an alternative to joint phrasal translation models. This syntactic model is similar to its flatstring phrasal predecessors, but admits polynomial-time algorithms for Viterbi alignment and EM training. We demonstrate that the consistency constraints that allow flat phrasal models to scale also help ITG algorithms, producing an 80-times faste...

KE N E W A RD One of the fi rst and most enduring facts most students learn in biology class is that all living cells use a small molecule called adenosine triphosphate (ATP) as fuel. That universal energy currency drives the biological reactions that allow cells to function and life to fl ourish—making ATP a crucial player in the biological world. Less commonly known, however, is that what is ...