Mirror-Generative Neural Machine Translation

ICLR, 2020.

Cited by: 11|Views373
EI
Weibo:
We propose the mirror-generative neural machine translation, a novel deep generative model which simultaneously models a pair of src2tgt and tgt2src translation models, as well as a pair of source and target language models, in a highly integrated way with the mirror property

Abstract:

Training neural machine translation models (NMT) requires a large amount of parallel corpus, which is scarce for many language pairs. However, raw non-parallel corpora are often easy to obtain. Existing approaches have not exploited the full potential of non-parallel bilingual data either in training or decoding. In this paper, we ...More

Code:

Data:

Your rating :
0

 

Tags
Comments