Inductive Subgraphs as Shortcuts: Causal Disentanglement for Heterophilic Graph Learning

2026-04-21

AI Summary

This work addresses the challenge of spurious shortcuts in heterophilous graphs, where inductive subgraphs often mislead graph neural networks (GNNs) by introducing non-causal correlations. From a causal inference perspective, the study is the first to formally characterize this mechanism and proposes a debiased causal graph that blocks confounding and spillover pathways. Building upon this, the authors introduce a causally disentangled GNN framework that effectively separates spurious associations from genuine causal subgraph signals. By innovatively integrating causal disentanglement into the elimination of non-causal paths, the method significantly enhances both accuracy and robustness in node classification. Extensive experiments demonstrate its consistent superiority over state-of-the-art heterophilous graph learning approaches across multiple real-world datasets.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Heterophily is a prevalent property of real-world graphs and is well known to impair the performance of homophilic Graph Neural Networks (GNNs). Prior work has attempted to adapt GNNs to heterophilic graphs through non-local neighbor extension or architecture refinement. However, the fundamental reasons behind misclassifications remain poorly understood. In this work, we take a novel perspective by examining recurring inductive subgraphs, empirically and theoretically showing that they act as spurious shortcuts that mislead GNNs and reinforce non-causal correlations in heterophilic graphs. To address this, we adopt a causal inference perspective to analyze and correct the biased learning behavior induced by shortcut inductive subgraphs. We propose a debiased causal graph that explicitly blocks confounding and spillover paths responsible for these shortcuts. Guided by this causal graph, we introduce Causal Disentangled GNN (CD-GNN), a principled framework that disentangles spurious inductive subgraphs from true causal subgraphs by explicitly blocking non-causal paths. By focusing on genuine causal signals, CD-GNN substantially improves the robustness and accuracy of node classification in heterophilic graphs. Extensive experiments on real-world datasets not only validate our theoretical findings but also demonstrate that our proposed CD-GNN outperforms state-of-the-art heterophily-aware baselines.

Problem

• heterophily
• inductive subgraphs
• spurious shortcuts
• causal disentanglement
• graph neural networks

Innovation

• causal disentanglement
• heterophilic graphs
• inductive subgraphs
• spurious shortcuts
• graph neural networks