In this paper, we enhance the attention-based neural machine translation (NMT) by adding explicit coverage embedding models to alleviate issues of repeating and dropping translations in NMT. For each source word, our model starts with a full coverage embedding vector to track the coverage status, and then keeps updating it with neural networks as the translation goes. Experiments on the large-scale Chinese-to-English task show that our enhanced model improves the translation quality significantly on various test sets over the strong large vocabulary NMT system.
@article{arxiv.1605.03148,
title = {Coverage Embedding Models for Neural Machine Translation},
author = {Haitao Mi and Baskaran Sankaran and Zhiguo Wang and Abe Ittycheriah},
journal= {arXiv preprint arXiv:1605.03148},
year = {2016}
}