arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1808.04339cs.CL
收藏

非平行语篇风格迁移的解纠缠表示学习

Disentangled Representation Learning for Non-Parallel Text Style Transfer

University of Waterloo(滑铁卢大学);AdeptMind Research(AdeptMind研究院)

Vineet John, Lili Mou, Hareesh Bahuleyan, Olga Vechtomova

摘要:

本文解决了语言模型中风格和内容潜变量的分离问题。我们提出了一种简单而有效的方法,它结合了辅助多任务和对抗目标,分别用于标签预测和词袋预测。我们表明,无论是定性和定量的,风格和内容确实是在潜在的空间解开。将这种分离的潜在表示学习方法应用于非平行语料库上的风格迁移。与以前的最先进方法相比,我们在传输准确性,内容保存和语言流畅性方面取得了更好的结果。

英文摘要:

This paper tackles the problem of disentangling the latent variables of style and content in language models. We propose a simple yet effective approach, which incorporates auxiliary multi-task and adversarial objectives, for label prediction and bag-of-words prediction, respectively. We show, both qualitatively and quantitatively, that the style and content are indeed disentangled in the latent space. This disentangled latent representation learning method is applied to style transfer on non-parallel corpora. We achieve substantially better results in terms of transfer accuracy, content preservation and language fluency, in comparison to previous state-of-the-art approaches.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑