Effectively training neural machine translation with monolingual data

CASIA OpenIR > 数字内容技术与服务研究中心 > 听觉模型与认知计算

	Effectively training neural machine translation with monolingual data
	Yang Z(杨振); Chen W(陈炜); Feng W（王峰）
发表期刊	NeuroComputing
	2018
期号	100 页码:10-20
摘要	Improving neural machine translation models (NMT) with monolingual data has aroused more and more interests in this area and back-translation for monolingual data augmentation \cite{sennrich2015improving} has been taken as a promising development recently. While the naive back-translation approach improves the translation performance substantially, we notice that its usage for monolingual data is not so effective because traditional NMT models make no distinction between the true parallel corpus and the back translated synthetic parallel corpus. This paper proposes a \textbf{\emph{gate-enhanced}} NMT model which makes use of monolingual data more effectively. The central idea is to separate the data flow of monolingual data and parallel data into different channels by the elegant designed gate, which enables the model to perform different transformations according to the type of the input sequence, i.e., monolingual data and parallel data. Experiments on Chinese-English and English-German translation tasks show that our approach achieves substantial improvements over strong baselines and the \textbf{\emph{gate-enhanced}} NMT model can utilize the source-side and target-side monolingual data at the same time.
关键词	Neural Machine Translation
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/22103
专题	数字内容技术与服务研究中心_听觉模型与认知计算
推荐引用方式 GB/T 7714	Yang Z,Chen W,Feng W. Effectively training neural machine translation with monolingual data[J]. NeuroComputing,2018(100):10-20.
APA	Yang Z,Chen W,&Feng W.(2018).Effectively training neural machine translation with monolingual data.NeuroComputing(100),10-20.
MLA	Yang Z,et al."Effectively training neural machine translation with monolingual data".NeuroComputing .100(2018):10-20.

条目包含的文件		下载所有文件
文件名称/大小	文献类型	版本类型	开放类型	使用许可
elsarticle-template.（491KB）	期刊论文	作者接受稿	开放获取	CC BY-NC-SA	浏览下载