Tag
Proposes a finite-sample method for recovering the sparsest DAG in linear non-Gaussian acyclic models with latent confounders using higher-order cumulants, without restricting the number of latents.