arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2107.02418cs.CL
收藏

自然证明生成的概率图推理

Probabilistic Graph Reasoning for Natural Proof Generation

ByteDance AI Lab(字节跳动人工智能实验室);Fudan University(复旦大学);University of Wisconsin–Madison(威斯康星大学麦迪逊分校);East China Normal University(华东师范大学)

Changzhi Sun, Xinbo Zhang, Jiangjie Chen, Chun Gan, Yuanbin Wu, Jiaze Chen, Hao Zhou, Lei Li

摘要:

在本文中,我们研究了基于自然语言语句的推理问题。先前的基于神经的方法没有明确考虑答案及其证明之间的相互依赖性。在本文中,我们提出了PRobr,联合答案预测和证明生成的一种新方法。PRobr定义了一个联合概率分布在所有可能的证明图和答案通过一个诱导的图形模型。然后,我们使用神经文本表示上的变分近似优化模型。在不同设置(完全监督、Few-Shot和zero-shot评估)下对多个数据集的实验验证了PRobr的有效性,例如,在少量/zero-shot评估中,QA准确度提高了10%-30%。我们的代码和模型可以在https://github.com/changzhisun/PRobr/上找到。

英文摘要:

In this paper, we investigate the problem of reasoning over natural language statements. Prior neural based approaches do not explicitly consider the inter-dependency among answers and their proofs. In this paper, we propose PRobr, a novel approach for joint answer prediction and proof generation. PRobr defines a joint probabilistic distribution over all possible proof graphs and answers via an induced graphical model. We then optimize the model using variational approximation on top of neural textual representation. Experiments on multiple datasets under diverse settings (fully supervised, few-shot and zero-shot evaluation) verify the effectiveness of PRobr, e.g., achieving 10%-30% improvement on QA accuracy in few/zero-shot evaluation. Our codes and models can be found at https://github.com/changzhisun/PRobr/.

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。

↑