arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.01454cs.LGcs.AIstat.MEstat.ML
收藏

将大型语言模型融入因果发现:一种统计因果方法

Integrating Large Language Models in Causal Discovery: A Statistical Causal Approach

Data Science and AI Innovation Research Promotion Center, Shiga University(Shiga大学数据科学与人工智能创新研究促进中心);National Institute of Science and Technology Policy (NISTEP)(国家科学技术政策研究所);Department of Health Data Science, Tokyo Medical University Graduate School of Medicine(东京医科大学医学研究生院健康数据科学系);Human Health Sciences, Kyoto University(京都大学人类健康科学);Graduate School of Medicine Human Health Sciences, Kyoto University(京都大学医学研究生院人类健康科学);Center for Advanced Intelligence Project, RIKEN(RIKEN高级智能项目);Department of Epidemiology Infectious Disease Control and Prevention, Hiroshima University(广岛大学流行病学与传染病防控系);Graduate School of Biomedical and Health Sciences Data Science and AI Innovation Research Promotion Center, Shiga University(Shiga大学生物医学与健康科学研究生院数据科学与人工智能创新研究促进中心);SANKEN, The University of Osaka Faculty of Data Science, Shiga University(大阪大学数据科学学院,Shiga大学SANKEN);Institute for the Advanced Study of Human Biology, Kyoto University(京都大学人类生物学高级研究机构);Department of Physics, Kyoto University(京都大学物理系);Graduate School of Engineering, The University of Tokyo(东京大学工学研究生院);Research and Development Center for Large Language Models, National Institute of Informatics(信息处理研究所大型语言模型研究开发中心)

Masayuki Takayama, Tadahisa Okuda, Thong Pham, Tatsuyoshi Ikenoue, Shingo Fukuma, Shohei Shimizu, Akiyoshi Sannai

AI总结:

本文提出统计因果提示与先验知识增强框架,将 LLM-KBCI 与 SCD 结合,证明 LLM 背景知识可提升因果发现效果。

摘要:

在实际的统计因果发现(SCD)中,将领域专家知识作为约束嵌入算法,对于获得能够反映领域专家广泛知识的合理因果模型十分重要,尽管系统化获取背景知识仍存在挑战。为克服这些挑战,本文提出一种新的因果推断方法,通过面向 LLM 的“统计因果提示(SCP)”以及面向 SCD 的先验知识增强,将 SCD 与基于大型语言模型(LLM)的知识型因果推断(KBCI)综合起来。本研究中的实验表明,LLM-KBCI 的结果以及经 LLM-KBCI 增强的 SCD 结果比没有先验知识的 SCD 结果更接近真实情况。实验还表明,如果 LLM 接受 SCP,SCD 结果可以进一步改善。此外,利用一个尚未公开的真实世界数据集,我们证明了 LLM 提供的背景知识能够改善该数据集上的 SCD,即使该数据集从未包含在 LLM 的训练数据中。为了未来将所提方法实际应用于医疗保健等重要领域,我们还深入讨论了其局限性、严重错误风险、围绕 LLM 的技术预期改进,以及将专家对结果的检查现实地整合到这一自动流程中的方式,并在成功和失败场景下的多种条件中进行了 SCP 模拟。因此,在各领域中对本文所提方法进行谨慎而适当的应用、改进和定制,可以应对数据集偏差与局限等挑战,展示 LLM 在不同科学领域改善数据驱动因果推断的潜力。本研究使用的代码公开发布于:www.github.com/mas-takayama/LLM-and-SCD。

英文摘要:

In practical statistical causal discovery (SCD), embedding domain expert knowledge as constraints into the algorithm is important for reasonable causal models reflecting the broad knowledge of domain experts, despite the challenges in the systematic acquisition of background knowledge. To overcome these challenges, this paper proposes a novel method for causal inference, in which SCD and knowledge-based causal inference (KBCI) with a large language model (LLM) are synthesized through ``statistical causal prompting (SCP)'' for LLMs and prior knowledge augmentation for SCD. The experiments in this work have revealed that the results of LLM-KBCI and SCD augmented with LLM-KBCI approach the ground truths, more than the SCD result without prior knowledge. These experiments have also revealed that the SCD result can be further improved if the LLM undergoes SCP. Furthermore, with an unpublished real-world dataset, we have demonstrated that the background knowledge provided by the LLM can improve the SCD on this dataset, even if this dataset has never been included in the training data of the LLM. For future practical application of this proposed method across important domains such as healthcare, we also thoroughly discuss the limitations, risks of critical errors, expected improvement of techniques around LLMs, and realistic integration of expert checks of the results into this automatic process, with SCP simulations under various conditions both in successful and failure scenarios. The careful and appropriate application of the proposed approach in this work, with improvement and customization for each domain, can thus address challenges such as dataset biases and limitations, illustrating the potential of LLMs to improve data-driven causal inference across diverse scientific domains. The code used in this work is publicly available at: www.github.com/mas-takayama/LLM-and-SCD

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。

↑