Speculative Beam Search for Simultaneous Translation
Speculative Beam Search for Simultaneous Translation
复制标题
用于同步翻译的推测波束搜索
DOI：
10.18653/v1/d19-1144
复制
发表时间：
2019
影响因子：
5.8
通讯作者：
Liang Huang
中科院分区：
文献类型：
--
作者：
Renjie Zheng;Mingbo Ma;Baigong Zheng;Liang Huang
Beam搜索广泛用于（整句）机器翻译，但其在同声翻译中的应用仍然非常重要，其中输出单词是动态提交的。特别是，最近提出的等待k政策（Ma等人，2018）是一种简单而有效的方法，（在初始等待之后）在接收到每个输入字时提交一个输出字，使得波束搜索看起来不适用。为了应对这一挑战，我们提出了一种新的推测性波束搜索算法，该算法可以在未来的几个步骤中产生幻觉，以便通过隐含地受益于目标语言模型来达到更准确的决策。这个想法使得波束搜索第一次适用于在每一步中生成单个单词。不同的语言对的实验表明，与以前的工作相比，有很大的改善。
Beam search is universally used in (full-sentence) machine translation but its application to simultaneous translation remains highly non-trivial, where output words are committed on the fly. In particular, the recently proposed wait-k policy (Ma et al., 2018) is a simple and effective method that (after an initial wait) commits one output word on receiving each input word, making beam search seemingly inapplicable. To address this challenge, we propose a new speculative beam search algorithm that hallucinates several steps into the future in order to reach a more accurate decision by implicitly benefiting from a target language model. This idea makes beam search applicable for the first time to the generation of a single word in each step. Experiments over diverse language pairs show large improvement compared to previous work.