Abstract:This article addresses two important issues of causal inference in the high-dimensional situation. One is how to reduce redundant conditional independence (CI) tests, which heavily impact the efficiency and accuracy of existing constraint-based methods. Another is how to construct the true causal graph from a set of Markov equivalence classes returned by these methods. For the first issue, we design a recursive decomposition approach where the original data (a set of variables) are first decomposed into two small subsets, each of which is then recursively decomposed into two smaller subsets until none of these subsets can be decomposed further. Redundant CI tests can be reduced by inferring causalities from these subsets. The advantage of this decomposition scheme lies in two aspects: 1) it requires only low-order CI tests and 2) it does not violate $d$ -separation. The complete causality can be reconstructed by merging all the partial results of the subsets. For the second issue, we employ regression-based CI tests to check CIs in linear non-Gaussian additive noise cases, which can identify more causal directions by $x - E(x|Z) \_{}\!\!\_{}\!\!\!\!\!\shortparallel z$ (or $y - E(y|Z) \_{}\!\!\_{}\!\!\!\!\!\shortparallel z$ ). Consequently, causal direction learning is no longer limited by the number of returned $V$ -structures and consistent propagation. Extensive experiments show that the proposed method can not only substantially reduce redundant CI tests but also effectively distinguish the equivalence classes.

Structure and Parameter Learning for Causal Independence and Causal Interaction Models

Learning Cluster Causal Diagrams: an Information-Theoretic Approach

Learning causal structures based on markov equivalence class

A Meta-Learning Approach to Bayesian Causal Discovery

Amortized Inference for Causal Structure Learning

Learning causal structures using hidden compact representation

Invariance & Causal Representation Learning: Prospects and Limitations

Conditionally-additive-noise Models for Structure Learning

Learning Causal Structures Based on Divide and Conquer

Causal Modeling in Multi-Context Systems: Distinguishing Multiple Context-Specific Causal Graphs which Account for Observational Support

Causal models on probability spaces

Conditional learning through causal models

Kernel-based independence tests for causal structure learning on functional data

Bivariate Causal Discovery using Bayesian Model Selection

Learning and Testing Causal Models with Interventions

Dual Likelihood for Causal Inference under Structure Uncertainty

Learning Causal Abstractions of Linear Structural Causal Models

Discovering Mixtures of Structural Causal Models from Time Series Data

Causal Discovery in Linear Models with Unobserved Variables and Measurement Error

Can Large Language Models Learn Independent Causal Mechanisms?

Unifying Causal Representation Learning with the Invariance Principle