Interpretable and Generalizable Graph Learning via Stochastic Attention Mechanism
This addresses the need for stable and accurate interpretability in graph learning for scientific applications, offering a novel method that avoids spurious correlations.
The paper tackles the problem of interpretable graph learning by proposing Graph Stochastic Attention (GSAT), which uses stochastic attention to select task-relevant subgraphs for interpretation, achieving up to 20% improvement in interpretation AUC and 5% in prediction accuracy over state-of-the-art methods.
Interpretable graph learning is in need as many scientific applications depend on learning models to collect insights from graph-structured data. Previous works mostly focused on using post-hoc approaches to interpret pre-trained models (graph neural networks in particular). They argue against inherently interpretable models because the good interpretability of these models is often at the cost of their prediction accuracy. However, those post-hoc methods often fail to provide stable interpretation and may extract features that are spuriously correlated with the task. In this work, we address these issues by proposing Graph Stochastic Attention (GSAT). Derived from the information bottleneck principle, GSAT injects stochasticity to the attention weights to block the information from task-irrelevant graph components while learning stochasticity-reduced attention to select task-relevant subgraphs for interpretation. The selected subgraphs provably do not contain patterns that are spuriously correlated with the task under some assumptions. Extensive experiments on eight datasets show that GSAT outperforms the state-of-the-art methods by up to 20%$\uparrow$ in interpretation AUC and 5%$\uparrow$ in prediction accuracy. Our code is available at https://github.com/Graph-COM/GSAT.