跳到主要导航 跳到搜索 跳到主要内容

Smart-Start Decoding for Neural Machine Translation

  • Jian Yang
  • , Shuming Ma
  • , Dongdong Zhang
  • , Juncheng Wan
  • , Zhoujun Li*
  • , Ming Zhou
  • *此作品的通讯作者
  • Microsoft USA
  • Shanghai Jiao Tong University

科研成果: 书/报告/会议事项章节 › 会议稿件 › 同行评审

摘要

Most current neural machine translation models adopt a monotonic decoding order of either left-to-right or right-to-left. In this work, we propose a novel method that breaks up the limitation of these decoding orders, called Smart-Start decoding. More specifically, our method first predicts a median word. It starts to decode the words on the right side of the median word and then generates words on the left. We evaluate the proposed Smart-Start decoding method on three datasets. Experimental results show that the proposed method can significantly outperform strong baseline models.

源语言英语
主期刊名NAACL-HLT 2021 - 2021 Conference of the North American Chapter of the Association for Computational Linguistics
主期刊副标题Human Language Technologies, Proceedings of the Conference
出版商Association for Computational Linguistics (ACL)
页3982-3988
页数7
ISBN(电子版)9781954085466
DOI
出版状态已出版 - 2021
活动2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2021 - Virtual, Online
期限: 6 6月 2021 → 11 6月 2021

丛书

姓名NAACL-HLT 2021 - 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Proceedings of the Conference

会议

会议2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2021
市Virtual, Online
时期6/06/21 → 11/06/21

学术指纹

探究 'Smart-Start Decoding for Neural Machine Translation' 的科研主题。它们共同构成独一无二的学术指纹。

引用此