English

Explicit Cross-lingual Pre-training for Unsupervised Machine Translation

Computation and Language 2019-09-04 v1

Abstract

Pre-training has proven to be effective in unsupervised machine translation due to its ability to model deep context information in cross-lingual scenarios. However, the cross-lingual information obtained from shared BPE spaces is inexplicit and limited. In this paper, we propose a novel cross-lingual pre-training method for unsupervised machine translation by incorporating explicit cross-lingual training signals. Specifically, we first calculate cross-lingual n-gram embeddings and infer an n-gram translation table from them. With those n-gram translation pairs, we propose a new pre-training model called Cross-lingual Masked Language Model (CMLM), which randomly chooses source n-grams in the input text stream and predicts their translation candidates at each time step. Experiments show that our method can incorporate beneficial cross-lingual information into pre-trained models. Taking pre-trained CMLM models as the encoder and decoder, we significantly improve the performance of unsupervised machine translation.

Keywords

Cite

@article{arxiv.1909.00180,
  title  = {Explicit Cross-lingual Pre-training for Unsupervised Machine Translation},
  author = {Shuo Ren and Yu Wu and Shujie Liu and Ming Zhou and Shuai Ma},
  journal= {arXiv preprint arXiv:1909.00180},
  year   = {2019}
}

Comments

Accepted to EMNLP2019; 10 pages, 2 figures

R2 v1 2026-06-23T11:02:02.996Z