Abstract

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

 

Quick Read (beta)