Improved zero-shot neural machine translation via ignoring spurious correlations

65Citations
Citations of this article
192Readers
Mendeley users who have this article in their library.

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, naïve 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) back-translation. 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.

Cite

CITATION STYLE

APA

Gu, J., Wang, Y., Cho, K., & Li, V. O. K. (2020). Improved zero-shot neural machine translation via ignoring spurious correlations. In ACL 2019 - 57th Annual Meeting of the Association for Computational Linguistics, Proceedings of the Conference (pp. 1258–1268). Association for Computational Linguistics (ACL). https://doi.org/10.18653/v1/p19-1121

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free