Relating Graph Neural Networks to Structural Causal Models
Causality can be described in terms of a structural causal model (SCM) that carries information on the variables of interest and their mechanistic relations. For most processes of interest the underlying SCM will only be partially observable, thus causal inference tries leveraging the exposed. Graph neural networks (GNN) as universal approximators on structured input pose a viable candidate for causal learning, suggesting a tighter integration with SCM. To this effect we present a theoretical analysis from first principles that establishes a more general view on neural-causal models, revealing several novel connections between GNN and SCM. We establish a new model class for GNN-based causal inference that is necessary and sufficient for causal effect identification. Our empirical illustration on simulations and standard benchmarks validate our theoretical proofs.
Relating Graph Neural Networks to Structural Causal Models Matej Zečević MATEJ . ZECEVIC @ CS . TU - DARMSTADT. DE Computer Science Department, TU Darmstadt Devendra Singh Dhami DEVENDRA . DHAMI @ CS . TU - DARMSTADT. DE Computer Science Department, TU Darmstadt Petar Veličković…
saved by
related reading
- A Gentle Introduction to Graph Neural Networksdistill.pub
- Understanding Convolutions on Graphsdistill.pub
- Beyond Message Passing: a Physics-Inspired Paradigm for Graph Neural Networksthegradient.pub
- Circuit Tracing: Revealing Computational Graphs in Language Modelstransformer-circuits.pub
- Faithful, Interpretable Model Explanations via Causal Abstraction | SAIL Blogai.stanford.edu
- Deep learning of causal structures in high dimensions under data limitations | Nature Machine Intelligencenature.com
- Weakly supervised causal representation learningarxiv.org
- [2604.21691] There Will Be a Scientific Theory of Deep Learningarxiv.org
- Causal Autoregressive Flowsarxiv.org
- [2304.07638] Causal models in string diagramsarxiv.org
- Graph Convolutional Networks | Thomas Kipf | Google DeepMindtkipf.github.io
- [2508.11214] How Causal Abstraction Underpins Computational Explanationarxiv.org