Domanial and dimensional adversarial learning for emotion regression
Domanial and dimensional adversarial learning for emotion regression
复制标题
DOI：
10.1016/j.neucom.2020.09.036
复制
发表时间：
2021
期刊：
影响因子：
6
通讯作者：
Suyang Zhu;Shoushan Li;Guodong Zhou
中科院分区：
文献类型：
--
作者：
Suyang Zhu;Shoushan Li;Guodong Zhou
在本文中，我们通过对抗性学习来解决跨域多维情绪回归。这是通过适当地进行次元和领域对抗性学习来实现的。一方面，我们通过维度鉴别器在情绪维度之间进行对抗性学习，通过注意机制获得维度特定的特征，从而更好地确定维度情绪得分。另一方面，我们通过领域鉴别器在目标领域和多个源域之间进行对抗性学习，以便更好地利用源域的文本在目标领域进行回归模型训练。在EMOBANK语料库上的实验结果表明，在最新的基线上，我们提出的方法在跨域多维情感回归任务中取得了显著的改善。
In this paper, we address cross-domain multi-dimensional emotion regression through adversarial learning. This is done via a proper conduction of both dimensional and domanial adversarial learning. On the one hand, we conduct adversarial learning between emotion dimensions via a dimensional discriminator to achieve dimension-specific features through the attention mechanism for better determining dimensional emotion scores. On the other hand, we conduct adversarial learning between the target domain and multiple source domains via a domanial discriminator for better leveraging texts from source domains for regression model training in the target domain. Empirical evaluation on the EMOBANK corpus shows that our proposed approach achieves notable improvements inr-values in the cross-domain multi-dimensional emotion regression task over the state-of-the-art baselines.