PulseAugur
EN
LIVE 07:30:11

Minimum Bayes Risk decoding improves machine translation error detection

By PulseAugur Editorial · [1 sources] ·

Researchers have introduced Minimum Bayes Risk (MBR) decoding as an alternative to Maximum a Posteriori (MAP) decoding for error span detection in machine translation evaluation. This new method aims to improve the localization and severity labeling of translation errors by selecting candidate hypotheses based on their similarity to human annotations. Experiments on the WMT24 Metrics Shared Task demonstrated that MBR decoding significantly enhances span-level performance and matches or surpasses MAP decoding at system and sentence levels. To address computational costs, the researchers developed a method to distill MBR decisions into a model decoded via greedy search, thereby eliminating inference-time latency. AI

IMPACT Introduces a novel decoding technique that could improve the accuracy and efficiency of automated machine translation quality assessment.

RANK_REASON Academic paper detailing a new decoding method for machine translation evaluation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Minimum Bayes Risk decoding improves machine translation error detection

COVERAGE [1]

  1. arXiv cs.AI TIER_1 English(EN) · Boxuan Lyu, Haiyue Song, Hidetaka Kamigaito, Chenchen Ding, Hideki Tanaka, Masao Utiyama, Kotaro Funakoshi, Manabu Okumura ·

    Minimum Bayes Risk Decoding for Error Span Detection in Reference-Free Automatic Machine Translation Evaluation

    arXiv:2512.07540v4 Announce Type: replace-cross Abstract: Error Span Detection (ESD) extends automatic machine translation (MT) evaluation by localizing translation errors and labeling their severity. Current generative ESD methods typically use Maximum a Posteriori (MAP) decodin…

RELATED TOPICS