Centroid-Based Efficient Minimum Bayes Risk Decoding
This work addresses efficiency and performance bottlenecks in machine translation decoding, offering a practical improvement for NLP practitioners, though it is incremental as it builds on existing MBR methods.
The paper tackled the quadratic time complexity of Minimum Bayes Risk (MBR) decoding in machine translation by proposing centroid-based MBR (CBMBR) decoding, which clusters reference translations and uses centroids to compute scores, resulting in a 5.7x speed improvement and up to 0.5 COMET gain in translation quality across multiple WMT tasks.
Minimum Bayes risk (MBR) decoding achieved state-of-the-art translation performance by using COMET, a neural metric that has a high correlation with human evaluation. However, MBR decoding requires quadratic time since it computes the expected score between a translation hypothesis and all reference translations. We propose centroid-based MBR (CBMBR) decoding to improve the speed of MBR decoding. Our method clusters the reference translations in the feature space, and then calculates the score using the centroids of each cluster. The experimental results show that our CBMBR not only improved the decoding speed of the expected score calculation 5.7 times, but also outperformed vanilla MBR decoding in translation quality by up to 0.5 COMET in the WMT'22 En$\leftrightarrow$Ja, En$\leftrightarrow$De, En$\leftrightarrow$Zh, and WMT'23 En$\leftrightarrow$Ja translation tasks.