English

On the pitfalls of Gaussian likelihood scoring for causal discovery

Methodology 2023-05-04 v2

Abstract

We consider likelihood score-based methods for causal discovery in structural causal models. In particular, we focus on Gaussian scoring and analyze the effect of model misspecification in terms of non-Gaussian error distribution. We present a surprising negative result for Gaussian likelihood scoring in combination with nonparametric regression methods.

Keywords

Cite

@article{arxiv.2210.11104,
  title  = {On the pitfalls of Gaussian likelihood scoring for causal discovery},
  author = {Christoph Schultheiss and Peter Bühlmann},
  journal= {arXiv preprint arXiv:2210.11104},
  year   = {2023}
}