In this paper we describe a method for Minimum Bayes Risk decoding for speech recognition. This is a technique similar to Consensus a.k.a. Confusion Network Decoding, in which we ...
Minimum Error Rate Training (MERT) and Minimum Bayes-Risk (MBR) decoding are used in most current state-of-theart Statistical Machine Translation (SMT) systems. The algorithms wer...
Shankar Kumar, Wolfgang Macherey, Chris Dyer, Fran...
We present Minimum Bayes-Risk (MBR) decoding over translation lattices that compactly encode a huge number of translation hypotheses. We describe conditions on the loss function t...
Roy Tromble, Shankar Kumar, Franz Josef Och, Wolfg...
We propose a novel objective function for discriminatively tuning log-linear machine translation models. Our objective explicitly optimizes the BLEU score of expected n-gram count...
Minimum Error Rate Training (MERT) is an effective means to estimate the feature function weights of a linear model such that an automated evaluation criterion for measuring syste...
Wolfgang Macherey, Franz Josef Och, Ignacio Thayer...