Abstract
In this paper, we present Neural Phrase-based Machine Translation (NPMT). Our method explicitly models the phrase structures in output sequences using Sleep-WAke Networks (SWAN), a recently proposed segmentation-based sequence modeling method. To mitigate the monotonic alignment requirement of SWAN, we introduce a new layer to perform (soft) local reordering of input sequences. Different from existing neural machine translation (NMT) approaches, NPMT does not use attention-based decoding mechanisms. Instead, it directly outputs phrases in a sequential order and can decode in linear time. Our experiments show that NPMT achieves superior performances on IWSLT 2014 German-English/English-German and IWSLT 2015 English-Vietnamese machine translation tasks compared with strong NMT baselines. We also observe that our method produces meaningful phrases in output languages.
Abstract (translated)
在本文中，我们提出了基于神经短语的机器翻译（NPMT）。我们的方法使用Sleep-WAke Networks（SWAN）明确地模拟输出序列中的短语结构，SWAN是最近提出的基于分段的序列建模方法。为了减轻SWAN的单调对齐要求，我们引入了一个新层来执行输入序列的（软）局部重新排序。与现有的神经机器翻译（NMT）方法不同，NPMT不使用基于注意力的解码机制。相反，它按顺序直接输出短语，并可以线性时间解码。我们的实验表明，与强大的NMT基线相比，NPMT在IWSLT 2014德语 - 英语/英语 - 德语和IWSLT 2015英语 - 越南语机器翻译任务中表现出色。我们还观察到我们的方法在输出语言中产生有意义的短语。
URL
https://arxiv.org/abs/1706.05565
