Ensemble Sequence Level Training for Multimodal MT: OSU-Baidu WMT18 Multimodal Machine Translation System Report
Ensemble Sequence Level Training for Multimodal MT: OSU-Baidu WMT18 Multimodal Machine Translation System Report
复制标题
DOI：
10.18653/v1/w18-6443
复制
发表时间：
2018-08
期刊：
影响因子：
--
通讯作者：
Renjie Zheng;Yilin Yang;Mingbo Ma;Liang Huang
中科院分区：
文献类型：
--
作者：
Renjie Zheng;Yilin Yang;Mingbo Ma;Liang Huang
本文介绍了俄勒冈州立大学与百度研究院为2018年WMT多模态翻译共享任务联合开发的多模态机器翻译系统。在本文中，我们提出了一种简单的方法，即通过将图像特征输入到解码器端来融入图像信息。我们还探索了不同的序列级训练方法，包括调度采样和强化学习，这些方法带来了显著的改进。我们的系统整合了多个采用不同架构和训练方法的模型，并在三个子任务中取得了最佳成绩：任务1中的英德和英捷翻译，以及任务1B中的（英+德+法）到捷克语的翻译。
This paper describes multimodal machine translation systems developed jointly by Oregon State University and Baidu Research for WMT 2018 Shared Task on multimodal translation. In this paper, we introduce a simple approach to incorporate image information by feeding image features to the decoder side. We also explore different sequence level training methods including scheduled sampling and reinforcement learning which lead to substantial improvements. Our systems ensemble several models using different architectures and training methods and achieve the best performance for three subtasks: En-De and En-Cs in task 1 and (En+De+Fr)-Cs task 1B.