Multi-View Causal Discovery without Non-Gaussianity: Identifiability and Algorithms

arXiv:2502.20115 · cs.LG, stat.ML · Submitted 2025-02-27 · Read on arXiv

cs.LG, stat.ML

Submitted: 2025-02-27

Updated: 2026-08-31

Comments: Accepted at the 43rd International Conference on Machine Learning (ICML 2026)

Code: https://github.com/cabal-cmu/Feedback-Discovery

License: http://creativecommons.org/licenses/by/4.0/

The gist: Causal discovery is a difficult problem that typically relies on strong assumptions on the data-generating model, such as non-Gaussianity.

Terminology

Abstract

Causal discovery is a difficult problem that typically relies on strong assumptions on the data-generating model, such as non-Gaussianity. In practice, many modern applications provide multiple related views of the same system, which has rarely been considered for causal discovery. Here, we leverage this multi-view structure to achieve causal discovery with weak assumptions. We propose a multi-view linear Structural Equation Model (SEM) that extends the well-known framework of non-Gaussian disturbances by alternatively leveraging correlation over views. We prove the identifiability of the model for acyclic SEMs. Subsequently, we propose several multi-view causal discovery algorithms, inspired by single-view algorithms (DirectLiNGAM, PairwiseLiNGAM, and ICA-LiNGAM). The new methods are validated through simulations and applications on neuroimaging data, where they enable the estimation of causal graphs between brain regions.

Related papers