Improving the Naturalness and Diversity of Referring Expression Generation models using Minimum Risk Training
Improving the Naturalness and Diversity of Referring Expression Generation models using Minimum Risk Training
复制标题
DOI：
10.18653/v1/2020.inlg-1.7
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Nikolaos Panagiaris;E. Hart;Dimitra Gkatzia
中科院分区：
文献类型：
--
作者：
Nikolaos Panagiaris;E. Hart;Dimitra Gkatzia
本文考虑具有序列水平目标的神经参照表达式生成(REG)模型的优化问题。最近，强化学习(RL)技术被用于训练深度端到端系统，以直接优化序列级别的目标。然而，与真实语言训练相关的两个问题是：(1)有效地应用真实语言是具有挑战性的；(2)由于生成的单词分布不足、词汇量较小以及频繁出现的单词和短语的重复性，生成的句子缺乏多样性和自然性。为了缓解这些问题，我们提出了一种新的训练REG模型的策略，使用最小风险训练(MRT)和最大似然估计(MLE)，我们的方法比RL W.r.t自然度和输出的多样性要好。具体地说，我们的方法在两个数据集中实现了23%-57%的苹果酒得分增长。通过与不同REG模型的详细比较，进一步证明了该方法的稳健性。
In this paper we consider the problem of optimizing neural Referring Expression Generation (REG) models with sequence level objectives. Recently reinforcement learning (RL) techniques have been adopted to train deep end-to-end systems to directly optimize sequence-level objectives. However, there are two issues associated with RL training: (1) effectively applying RL is challenging, and (2) the generated sentences lack in diversity and naturalness due to deficiencies in the generated word distribution, smaller vocabulary size, and repetitiveness of frequent words and phrases. To alleviate these issues, we propose a novel strategy for training REG models, using minimum risk training (MRT) with maximum likelihood estimation (MLE) and we show that our approach outperforms RL w.r.t naturalness and diversity of the output. Specifically, our approach achieves an increase in CIDEr scores between 23%-57% in two datasets. We further demonstrate the robustness of the proposed method through a detailed comparison with different REG models.