arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2005.01279cs.CLcs.LG
收藏

通过模拟遥远的未来来改进对抗性文本生成

Improving Adversarial Text Generation by Modeling the Distant Future

Duke University(杜克大学);University at Buffalo(布法罗大学);Microsoft Dynamics 365 AI(微软Dynamics 365 AI);Citadel LLC(城堡投资有限责任公司);DeepMind(深度思维)

Ruiyi Zhang, Changyou Chen, Zhe Gan, Wenlin Wang, Dinghan Shen, Guoyin Wang, Zheng Wen, Lawrence Carin

摘要:

自回归文本生成模型通常关注局部流畅性,在长文本生成中可能导致语义不一致。此外,自动生成具有相似语义的单词是具有挑战性的,并且手工制作的语言规则难以应用。我们考虑了一个文本规划方案,并提出了一个基于模型的模仿学习方法来缓解上述问题。具体来说,我们提出了一种新的引导器网络,专注于更长时间范围内的生成过程,它可以帮助下一个词的预测,并为生成器优化提供中间奖励。大量的实验表明,所提出的方法导致改善的性能。

英文摘要:

Auto-regressive text generation models usually focus on local fluency, and may cause inconsistent semantic meaning in long text generation. Further, automatically generating words with similar semantics is challenging, and hand-crafted linguistic rules are difficult to apply. We consider a text planning scheme and present a model-based imitation-learning approach to alleviate the aforementioned issues. Specifically, we propose a novel guider network to focus on the generative process over a longer horizon, which can assist next-word prediction and provide intermediate rewards for generator optimization. Extensive experiments demonstrate that the proposed method leads to improved performance.

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。

↑