论文标题
通过惩罚低相互信息，专门对多域NMT进行专门域NMT
Specializing Multi-domain NMT via Penalizing Low Mutual Information
论文作者
论文摘要
多域神经机器翻译（NMT）训练具有多个域的单个模型。由于它在一个模型中处理多个领域的功效，因此它具有吸引力。理想的多域NMT应该同时学习独特的领域特征，但是，掌握域特殊性是一项非平凡的任务。在本文中，我们通过相互信息（MI）的角度研究了特定于域的信息，并提出了一个新的目标，该目标使低MI变得更高。我们的方法在当前竞争性的多域NMT模型中实现了最先进的性能。同样，我们从经验上表明，我们的目标促进了低MI，导致了域特有的多域NMT。
Multi-domain Neural Machine Translation (NMT) trains a single model with multiple domains. It is appealing because of its efficacy in handling multiple domains within one model. An ideal multi-domain NMT should learn distinctive domain characteristics simultaneously, however, grasping the domain peculiarity is a non-trivial task. In this paper, we investigate domain-specific information through the lens of mutual information (MI) and propose a new objective that penalizes low MI to become higher. Our method achieved the state-of-the-art performance among the current competitive multi-domain NMT models. Also, we empirically show our objective promotes low MI to be higher resulting in domain-specialized multi-domain NMT.
