针对条件 IV 条件集依赖完整 DAG、难以从数据发现的问题,本文基于 MAG 建立祖先 IV 理论并提出无偏因果效应估计算法,实验验证其优于现有 IV 方法。
摘要:
未观测混杂是从观测数据估计因果效应的主要障碍。当存在潜在混杂因素时,工具变量(IV)被广泛用于因果效应估计。在标准 IV 方法中,当给定 IV 有效时,可以获得无偏估计,但对标准 IV 的有效性要求严格且不可检验。条件 IV 已被提出,通过以一组观测变量(称为条件 IV 的条件集)为条件来放宽标准 IV 的要求。然而,为条件 IV 寻找条件集的准则需要一个表示观测变量和未观测变量之间因果关系的有向无环图(DAG)。这使得直接从数据中发现条件集具有挑战性。本文利用最大祖先图(MAG)进行含潜在变量的因果推断,研究了祖先 IV(一种使用 MAG 的条件 IV)的图性质,并发展了相关理论,以支持在预处理变量假设下从数据中为给定祖先 IV 数据驱动地发现条件集。基于该理论,我们开发了一种算法,利用给定祖先 IV 和观测数据进行无偏因果效应估计。在合成数据集和真实数据集上的大量实验证明了该算法相较于现有 IV 方法的性能。
英文摘要:
Unobserved confounding is the main obstacle to causal effect estimation from observational data. Instrumental variables (IVs) are widely used for causal effect estimation when there exist latent confounders. With the standard IV method, when a given IV is valid, unbiased estimation can be obtained, but the validity requirement on a standard IV is strict and untestable. Conditional IVs have been proposed to relax the requirement of standard IVs by conditioning on a set of observed variables (known as a conditioning set for a conditional IV). However, the criterion for finding a conditioning set for a conditional IV needs a directed acyclic graph (DAG) representing the causal relationships of both observed and unobserved variables. This makes it challenging to discover a conditioning set directly from data. In this paper, by leveraging maximal ancestral graphs (MAGs) for causal inference with latent variables, we study the graphical properties of ancestral IVs, a type of conditional IVs using MAGs, and develop the theory to support data-driven discovery of the conditioning set for a given ancestral IV in data under the pretreatment variable assumption. Based on the theory, we develop an algorithm for unbiased causal effect estimation with a given ancestral IV and observational data. Extensive experiments on synthetic and real-world datasets demonstrate the performance of the algorithm in comparison with existing IV methods.
Root Cause Analysis with Latent Confounders using Partial Ancestral Graphs
Signal Processing Laboratory, São Carlos School of Engineering, University of São Paulo(圣保罗大学圣卡洛斯工程学院信号处理实验室);Guaratinguetá School of Engineering, São Paulo State University(圣保罗州立大学瓜拉廷加工程学院)
Constraint-Based Causal Discovery using Partial Ancestral Graphs in the presence of Cycles
Korteweg-de Vries Institute(科特韦格-德弗里斯研究所);University of Amsterdam(阿姆斯特丹大学);Institute for Computing and Information Sciences(计算与信息科学研究所);Radboud University Nijmegen(拉德堡德大学奈梅亨分校)
IV Co-Scientist: Multi-Agent LLM Framework for Causal Instrumental Variable Discovery
CISPA Helmholtz Center for Information Security(CISPA亥姆霍兹信息安全中心);Max Planck Institute for Intelligent Systems(马克斯·普朗克智能系统研究所);Jinesis AI Lab(Jinesis AI实验室);University of Toronto(多伦多大学);Vector Institute(向量研究所);EuroSafeAI;Carnegie Mellon University(卡内基梅隆大学);Amazon(亚马逊)
Disentangled Instrumental Variables for Causal Inference with Networked Observational Data
School of Computer Science and Engineering(计算机科学与工程学院);Guangxi Key Lab of MSMI(广西 MSMI 重点实验室);MOE Key Lab of EBIT, Guangxi Normal University(教育部 EBIT 重点实验室,广西师范大学);School of Computer Science and Technology(计算机科学与技术学院);University of South Australia(南澳大利亚大学);School of Information Technology and Mathematical Sciences(信息科技与数学科学学院);China University of Mining and Technology(中国矿业大学)
Signature Kernel Conditional Independence Tests in Causal Discovery for Stochastic Processes
Technical University of Munich(慕尼黑工业大学);Helmholtz Munich(慕尼黑亥姆霍兹中心);Munich Center for Machine Learning(慕尼黑机器学习中心);University of Oxford(牛津大学);Lund University(隆德大学);Imperial College London(伦敦帝国学院)
Causal Foundation Models: Disentangling Physics from Instrument Properties
Massachusetts Institute of Technology(麻省理工学院);Harvard University(哈佛大学);New York University(纽约大学);Flatiron Institute(Flatiron研究所);The NSF AI Institute for Artificial Intelligence and Fundamental Interactions(NSF人工智能与基本相互作用研究所)