Improved Zero-shot Neural Machine Translation via Ignoring Spurious Correlations

Jiatao Gu^, Yong Wang^, Kyunghyun Cho, Victor O.K. Li

The 57th Annual Meeting of the Association for Computational Linguistics (ACL), 2019


png Abstract
Zero-shot translation, translating between language pairs on which a Neural Machine Translation (NMT) system has never been trained, is an emergent property when training the system in multilingual settings. However, na¨ıve training for zero-shot NMT easily fails, and is sensitive to hyper-parameter setting. The performance typically lags far behind the more conventional pivot-based approach which translates twice using a third language as a pivot. In this work, we address the degeneracy problem due to capturing spurious correlations by quantitatively analyzing the mutual information between language IDs of the source and decoded sentences. Inspired by this analysis, we propose to use two simple but effective approaches: (1) decoder pre-training; (2) backtranslation. These methods show significant improvement (4 ∼ 22 BLEU points) over the vanilla zero-shot translation on three challenging multilingual datasets, and achieve similar or better results than the pivot-based approach.

[arxiv] [code]

Please cite as:

@inproceedings{gu-etal-2019-improved,
    title = "Improved Zero-shot Neural Machine Translation via Ignoring Spurious Correlations",
    author = "Gu, Jiatao  and Wang, Yong  and Cho, Kyunghyun  and Li, Victor O.K.",
    booktitle = "Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics",
    month = jul,
    year = "2019",
    address = "Florence, Italy",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/P19-1121",
    pages = "1258--1268"
}