Identifiable Multi-View Causal Discovery Without Non-Gaussianity

2025-02-27

AI Summary

Linear causal discovery in multi-view structural equation models (SEM) suffers from parameter unidentifiability without non-Gaussian noise assumptions. Method: We propose a novel identifiability condition based on cross-view perturbation variance diversity, integrating multi-view SEM modeling, multi-view independent component analysis (ICA), and variance-difference-driven causal parameter estimation. Contribution/Results: For the first time, we rigorously prove full-parameter identifiability of multi-view SEM under only two conditions: (i) heterogeneity in perturbation variances across views and (ii) acyclicity of the underlying directed acyclic graph (DAG). Extensive experiments on synthetic data and real neuroimaging datasets demonstrate successful reconstruction of directed causal graphs among brain regions, validating the method’s effectiveness, robustness, and generalizability. This work breaks the long-standing reliance on non-Gaussian noise in linear causal discovery and establishes a new paradigm for causal inference from heterogeneous multi-source observations.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Reasoning under Uncertainty: Causality
• Computer Vision: Multi-modal Vision

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

We propose a novel approach to linear causal discovery in the framework of multi-view Structural Equation Models (SEM). Our proposed model relaxes the well-known assumption of non-Gaussian disturbances by alternatively assuming diversity of variances over views, making it more broadly applicable. We prove the identifiability of all the parameters of the model without any further assumptions on the structure of the SEM other than it being acyclic. We further propose an estimation algorithm based on recent advances in multi-view Independent Component Analysis (ICA). The proposed methodology is validated through simulations and application on real neuroimaging data, where it enables the estimation of causal graphs between brain regions.

Problem

• Identifies causal relationships in multi-view SEM
• Relaxes non-Gaussianity assumption using variance diversity
• Validates method with neuroimaging data analysis

Innovation

• Multi-view SEM for causal discovery
• Relaxed non-Gaussianity by variance diversity
• ICA-based algorithm for parameter estimation