UCAM Biomedical translation at WMT19: Transfer learning multi-domain ensembles

2019/06/13 by Danielle Saunders, Saunders, Danielle, Felix Stahlberg +3
Computer Science · #Computation and Language (cs.CL) #FOS: Computer and information sciences #Natural Language Processing Techniques #Text Readability and Simplification #Topic Modeling #cs.CL

paper · pdf · doi:10.48550/arxiv.1906.05786

To appear at WMT19

arxiv created 2019/06/13 · openalex publication_date 2019/06/13 · arxiv updated 2019/06/14 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

The 2019 WMT Biomedical translation task involved translating Medline abstracts. We approached this using transfer learning to obtain a series of strong neural models on distinct domains, and combining them into multi-domain ensembles. We further experiment with an adaptive language-model ensemble weighting scheme. Our submission achieved the best submitted results on both directions of English-Spanish.

Related

• Neural Machine Translation of Clinical Text: An Empirical Investigation into Multilingual Pre-Trained Language Models and Transfer-Learning
• Investigating Massive Multilingual Pre-Trained Machine Translation Models for Clinical Domain via Transfer Learning
• UFRGS Participation on the WMT Biomedical Translation Shared Task
• The University of Edinburgh's Neural MT Systems for WMT17
• Improving Similar Language Translation With Transfer Learning