From Bilingual to Multilingual Neural Machine Translation by Incremental\n Training

2019/06/28 by Carlos Escolano, Escolano, Carlos, Marta R. Costa‐jussà +3
Computer Science · #Computation and Language (cs.CL) #FOS: Computer and information sciences #Multimodal Machine Learning Applications #Natural Language Processing Techniques #Topic Modeling

paper · pdf · doi:10.48550/arxiv.1907.00735

openalex publication_date 2019/06/28 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Multilingual Neural Machine Translation approaches are based on the use of\ntask-specific models and the addition of one more language can only be done by\nretraining the whole system. In this work, we propose a new training schedule\nthat allows the system to scale to more languages without modification of the\nprevious components based on joint training and language-independent\nencoder/decoder modules allowing for zero-shot translation. This work in\nprogress shows close results to the state-of-the-art in the WMT task.\n

Related

• Building Multilingual Machine Translation Systems That Serve Arbitrary X-Y Translations
• Massively Multilingual Neural Machine Translation in the Wild: Findings and Challenges
• Multi-Way, Multilingual Neural Machine Translation with a Shared Attention Mechanism
• A Study of Multilingual Neural Machine Translation
• Improving Massively Multilingual Neural Machine Translation and Zero-Shot Translation