An Empirical Investigation of Multi-bridge Multilingual NMT models
This addresses translation efficiency and quality for low-resource languages by enabling a single multilingual system to handle all directions, though it is incremental on existing multilingual NMT work.
The paper investigates multi-bridge multilingual neural machine translation models trained on non-English language pairs, finding that they can overcome zero-shot translation issues, are competitive with pivot models using limited data, and can match or outperform English-centric models to serve all translation directions in a single system.
In this paper, we present an extensive investigation of multi-bridge, many-to-many multilingual NMT models (MB-M2M) ie., models trained on non-English language pairs in addition to English-centric language pairs. In addition to validating previous work which shows that MB-M2M models can overcome zeroshot translation problems, our analysis reveals the following results about multibridge models: (1) it is possible to extract a reasonable amount of parallel corpora between non-English languages for low-resource languages (2) with limited non-English centric data, MB-M2M models are competitive with or outperform pivot models, (3) MB-M2M models can outperform English-Any models and perform at par with Any-English models, so a single multilingual NMT system can serve all translation directions.