The paper proposes a contrastive-learning-based causal graph regression framework that explicitly models the predictive power of confounding subgraphs and achieves state-of-the-art OOD generalization on graph regression benchmarks.
Debiasing graph neural networks via learning disentangled causal substructure
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
A Recipe for Causal Graph Regression: Confounding Effects Revisited
The paper proposes a contrastive-learning-based causal graph regression framework that explicitly models the predictive power of confounding subgraphs and achieves state-of-the-art OOD generalization on graph regression benchmarks.