Adversarial Neural Machine Translation
Adversarial Neural Machine Translation
复制标题
DOI：
--
复制
发表时间：
2017-04
影响因子：
5.2
通讯作者：
Lijun Wu;Yingce Xia;Li Zhao;Fei Tian;Tao Qin;J. Lai;Tie-Yan Liu
中科院分区：
文献类型：
--
作者：
Lijun Wu;Yingce Xia;Li Zhao;Fei Tian;Tao Qin;J. Lai;Tie-Yan Liu
在本文中，我们研究了一种新的学习范式神经机器翻译（NMT）。我们不是像以前的作品那样最大化人工翻译的可能性，而是最小化人工翻译和NMT模型给出的翻译之间的区别。为了实现这一目标，受最近生成对抗网络（GAN）成功的启发，我们采用了一种对抗训练架构，并将其命名为对抗NMT。在Adversarial-NMT中，NMT模型的训练由对手辅助，对手是精心设计的卷积神经网络（CNN）。对手的目标是区分NMT模型生成的翻译结果与人类的翻译结果。NMT模型的目标是产生高质量的翻译，以欺骗对手。利用策略梯度方法来共同训练NMT模型和对手。在英语$\rightarrow$法语和德语$\rightarrow$英语翻译任务上的实验结果表明，Adversarial-NMT比几种强基线能显著提高翻译质量。
In this paper, we study a new learning paradigm for Neural Machine Translation (NMT). Instead of maximizing the likelihood of the human translation as in previous works, we minimize the distinction between human translation and the translation given by an NMT model. To achieve this goal, inspired by the recent success of generative adversarial networks (GANs), we employ an adversarial training architecture and name it as Adversarial-NMT. In Adversarial-NMT, the training of the NMT model is assisted by an adversary, which is an elaborately designed Convolutional Neural Network (CNN). The goal of the adversary is to differentiate the translation result generated by the NMT model from that by human. The goal of the NMT model is to produce high quality translations so as to cheat the adversary. A policy gradient method is leveraged to co-train the NMT model and the adversary. Experimental results on English$\rightarrow$French and German$\rightarrow$English translation tasks show that Adversarial-NMT can achieve significantly better translation quality than several strong baselines.