Unsupervised Neural Machine Translation Initialized by Unsupervised Statistical Machine Translation

TL;DR

This paper introduces a novel approach to unsupervised neural machine translation by initializing training with synthetic bilingual data generated through unsupervised statistical machine translation, leading to improved performance.

Contribution

It presents a method that combines unsupervised statistical and neural machine translation, initializing NMT with synthetic data to enhance unsupervised translation quality.

Findings

Achieved state-of-the-art results on WMT16 German-English translation.

Demonstrated the effectiveness of synthetic data initialization in UNMT.

Improved translation performance in both directions.

Abstract

Recent work achieved remarkable results in training neural machine translation (NMT) systems in a fully unsupervised way, with new and dedicated architectures that rely on monolingual corpora only. In this work, we propose to define unsupervised NMT (UNMT) as NMT trained with the supervision of synthetic bilingual data. Our approach straightforwardly enables the use of state-of-the-art architectures proposed for supervised NMT by replacing human-made bilingual data with synthetic bilingual data for training. We propose to initialize the training of UNMT with synthetic bilingual data generated by unsupervised statistical machine translation (USMT). The UNMT system is then incrementally improved using back-translation. Our preliminary experiments show that our approach achieves a new state-of-the-art for unsupervised machine translation on the WMT16 German--English news translation task, for both translation directions.