Sampling-Based Approximations to Minimum Bayes Risk Decoding for Neural Machine Translation. (arXiv:2108.04718v2 [cs.CL] UPDATED)

In NMT we search for the mode of the model distribution to form predictions.
The mode and other high-probability translations found by beam search have been
shown to often be inadequate in a number of ways. This prevents improving
translation quality through better search, as these idiosyncratic translations
end up selected by the decoding algorithm, a problem known as the beam search
curse. Recently, an approximation to minimum Bayes risk (MBR) decoding has been
proposed as an alternative decision rule …

arxiv bayes machine machine translation neural machine translation risk sampling translation