arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1909.07547cs.LGcs.AIstat.ML
收藏

面向开放域对话的分层强化学习

Hierarchical Reinforcement Learning for Open-Domain Dialog

Harvard University(哈佛大学);MIT Media Lab(麻省理工学院媒体实验室)

Abdelrhman Saleh, Natasha Jaques, Asma Ghandeharioun, Judy Hanwen Shen, Rosalind Picard

摘要:

开放域对话生成是一个具有挑战性的问题;最大似然训练可能导致重复输出,模型难以跟踪长期对话目标,并且在标准电影或在线数据集上进行训练可能导致生成不适当的,有偏见的或冒犯性的文本。强化学习(RL)是一个强大的框架,可以潜在地解决这些问题,例如通过允许对话模型优化以减少毒性和重复性。然而,以前将RL应用于开放域对话生成的方法是在单词级别进行的,这使得模型难以学习适当的信用分配以获得长期的会话奖励。在本文中,我们提出了一种新的方法,分层强化学习,VHRL,它使用策略梯度调整的话语级嵌入的变分序列模型。这种分层方法为学习长期的会话奖励提供了更大的灵活性。我们使用自我发挥和强化学习来优化一组以人为中心的对话指标,并表明我们的方法提供了显着的改进-在人类评估和自动指标方面-超过最先进的对话模型,包括Transformers。

英文摘要:

Open-domain dialog generation is a challenging problem; maximum likelihood training can lead to repetitive outputs, models have difficulty tracking long-term conversational goals, and training on standard movie or online datasets may lead to the generation of inappropriate, biased, or offensive text. Reinforcement Learning (RL) is a powerful framework that could potentially address these issues, for example by allowing a dialog model to optimize for reducing toxicity and repetitiveness. However, previous approaches which apply RL to open-domain dialog generation do so at the word level, making it difficult for the model to learn proper credit assignment for long-term conversational rewards. In this paper, we propose a novel approach to hierarchical reinforcement learning, VHRL, which uses policy gradients to tune the utterance-level embedding of a variational sequence model. This hierarchical approach provides greater flexibility for learning long-term, conversational rewards. We use self-play and RL to optimize for a set of human-centered conversation metrics, and show that our approach provides significant improvements -- in terms of both human evaluation and automatic metrics -- over state-of-the-art dialog models, including Transformers.

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。

↑