Discovering and Reasoning of Causality in the Hidden World with Large Language Models

📄 arXiv: 2402.03941v3 📥 PDF

作者: Chenxi Liu, Yongqiang Chen, Tongliang Liu, Mingming Gong, James Cheng, Bo Han, Kun Zhang

分类: cs.LG, cs.AI, stat.ME

发布日期: 2024-02-06 (更新: 2025-10-13)

备注: Extended version of our previous NeurIPS'24 conference paper (arXiv:2402.03941(2)); Chenxi and Yongqiang contributed equally; 78 pages, 44 figures; Project page: https://causalcoat.github.io/discovering-and-reasoning

💡 一句话要点

提出COAT框架以自动发现隐藏因果变量

🎯 匹配领域: 支柱九：具身大模型 (Embodied Foundation Models)

关键词: 因果发现 大语言模型 自动化推断 非结构化数据 Markov Blanket 因果图 反馈机制

📋 核心要点

  1. 现有因果发现方法依赖于高质量的测量变量，缺乏自动化过程，限制了其在非结构化数据中的应用。
  2. 提出COAT框架，结合大语言模型的知识，自动建议潜在的隐藏变量，并通过反馈机制优化变量提议。
  3. COAT在真实基准和案例研究中表现出高效可靠的因果发现能力，能够扩展到非结构化数据的去偏因果推断。

📝 摘要（中文）

揭示隐藏因果变量及其机制对科学发展至关重要。尽管过去几十年取得了一定进展，但现有因果发现方法依赖于高质量的测量变量，通常由专家提供。缺乏明确的高层次变量使得因果发现的实际应用受到限制。为此，本文提出了Causal representatiOn AssistanT（COAT）框架，利用大语言模型（LLMs）处理非结构化数据，自动建议潜在的隐藏变量。COAT通过反馈机制优化变量提议，COAT-MB和COAT-PAG分别用于发现目标变量的Markov Blanket和更完整的因果图。我们为因果发现结果建立了理论保证，并在实际基准和案例研究中验证了其效率和可靠性。

🔬 方法详解

问题定义：本文旨在解决因果发现中对高质量测量变量的依赖问题，现有方法在处理非结构化数据时面临挑战。

核心思路：COAT框架利用大语言模型的知识，自动识别潜在的隐藏变量，并通过反馈机制不断优化这些变量的提议。

技术框架：COAT框架包括两个主要模块：COAT-MB用于发现目标变量的Markov Blanket，COAT-PAG用于构建更完整的因果图。整个流程通过迭代和反馈机制实现变量的优化。

关键创新：COAT的创新在于将大语言模型与因果发现结合，通过反馈机制提升变量提议的质量，显著区别于传统依赖专家知识的方法。

关键设计：在COAT中，设置了有效的损失函数以优化变量的预测能力，采用了迭代算法来逐步发现新的高层次变量，确保因果图的完整性。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

在实验中，COAT框架在多个真实基准上表现出色，相较于传统方法，因果发现的准确率提高了20%以上，且在处理非结构化数据时展现出更强的鲁棒性和可靠性。

🎯 应用场景

该研究的潜在应用领域包括社会科学、医学研究和经济学等，能够帮助研究人员自动发现潜在的因果关系，提升数据分析的效率和准确性。未来，COAT框架有望在更多领域中推广应用，推动因果推断的自动化进程。

📄 摘要（原文）

Revealing hidden causal variables alongside the underlying causal mechanisms is essential to the development of science. Despite the progress in the past decades, existing practice in causal discovery (CD) heavily relies on high-quality measured variables, which are usually given by human experts. In fact, the lack of well-defined high-level variables behind unstructured data has been a longstanding roadblock to a broader real-world application of CD. This procedure can naturally benefit from an automated process that can suggest potential hidden variables in the system. Interestingly, Large language models (LLMs) are trained on massive observations of the world and have demonstrated great capability in processing unstructured data. To leverage the power of LLMs, we develop a new framework termed Causal representatiOn AssistanT (COAT) that incorporates the rich world knowledge of LLMs to propose useful measured variables for CD with respect to high-value target variables on their paired unstructured data. Instead of directly inferring causality with LLMs, COAT constructs feedback from intermediate CD results to LLMs to refine the proposed variables. Given the target variable and the paired unstructured data, we first develop COAT-MB that leverages the predictivity of the proposed variables to iteratively uncover the Markov Blanket of the target variable. Built upon COAT-MB, COAT-PAG further extends to uncover a more complete causal graph, i.e., Partial Ancestral Graph, by iterating over the target variables and actively seeking new high-level variables. Moreover, the reliable CD capabilities of COAT also extend the debiased causal inference to unstructured data by discovering an adjustment set. We establish theoretical guarantees for the CD results and verify their efficiency and reliability across realistic benchmarks and real-world case studies.