Discovering Invariant Rationales for Graph Neural Networks

Abstract

Intrinsic interpretability of graph neural networks (GNNs) is to find a small subset of the input graph's features -- rationale -- which guides the model prediction. Unfortunately, the leading rationalization models often rely on data biases, especially shortcut features, to compose rationales and make predictions without probing the critical and causal patterns. Moreover, such data biases easily change outside the training distribution. As a result, these models suffer from a huge drop in interpretability and predictive performance on out-of-distribution data. In this work, we propose a new strategy of discovering invariant rationale (DIR) to construct intrinsically interpretable GNNs. It conducts interventions on the training distribution to create multiple interventional distributions. Then it approaches the causal rationales that are invariant across different distributions while filtering out the spurious patterns that are unstable. Experiments on both synthetic and real-world datasets validate the superiority of our DIR in terms of interpretability and generalization ability on graph classification over the leading baselines. Code and datasets are available at https://github.com/Wuyxin/DIR-GNN.

arXiv commentICLR 2022

Cited by in corpus (12)

• Causal Attention for Interpretable and Generalizable Graph Classification
• CrossCBR: Cross-view Contrastive Learning for Bundle Recommendation
• A Comprehensive Survey on Trustworthy Graph Neural Networks: Privacy, Robustness, Fairness, and Explainability
• Graph Transformer for Recommendation
• Graph Rationalization with Environment-based Augmentations
• Time-aware Path Reasoning on Knowledge Graph for Recommendation
• GIF: A General Graph Unlearning Strategy via Influence Function
• Causality-based CTR Prediction using Graph Neural Networks
• GALA: Graph Diffusion-based Alignment with Jigsaw for Source-free Domain Adaptation
• Redundancy-aware Transformer for Video Question Answering
• Skew Class-balanced Re-weighting for Unbiased Scene Graph Generation
• Bridged-GNN: Knowledge Bridge Learning for Effective Knowledge Transfer