From GNNs to Trees: Multi-Granular Interpretability for Graph Neural Networks

2025-05-01

AI Summary

Existing subgraph-based GNN explanation methods overemphasize local structures while neglecting long-range dependencies; graph coarsening approaches improve global interpretability but suffer from fixed granularity, limiting adaptability to multi-scale real-world tasks. To address this, we propose the Tree-shaped Interpretable Framework (TIF), the first method to construct a multi-granularity hierarchical tree: it iteratively applies graph coarsening and branch perturbation to generate cross-scale subgraph nodes, and introduces an adaptive routing mechanism to dynamically select critical reasoning paths. TIF breaks rigid granularity constraints, enabling synergistic interpretation of both local subgraphs and global topology. Evaluated on synthetic and real-world graph classification benchmarks, TIF significantly improves explanation quality—measured by faithfulness and compactness—while maintaining prediction accuracy comparable to state-of-the-art models.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Graphical Models
• Humans and AI: Explainable AI (XAI) for Human Understanding

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Interpretable Graph Neural Networks (GNNs) aim to reveal the underlying reasoning behind model predictions, attributing their decisions to specific subgraphs that are informative. However, existing subgraph-based interpretable methods suffer from an overemphasis on local structure, potentially overlooking long-range dependencies within the entire graphs. Although recent efforts that rely on graph coarsening have proven beneficial for global interpretability, they inevitably reduce the graphs to a fixed granularity. Such an inflexible way can only capture graph connectivity at a specific level, whereas real-world graph tasks often exhibit relationships at varying granularities (e.g., relevant interactions in proteins span from functional groups, to amino acids, and up to protein domains). In this paper, we introduce a novel Tree-like Interpretable Framework (TIF) for graph classification, where plain GNNs are transformed into hierarchical trees, with each level featuring coarsened graphs of different granularity as tree nodes. Specifically, TIF iteratively adopts a graph coarsening module to compress original graphs (i.e., root nodes of trees) into increasingly coarser ones (i.e., child nodes of trees), while preserving diversity among tree nodes within different branches through a dedicated graph perturbation module. Finally, we propose an adaptive routing module to identify the most informative root-to-leaf paths, providing not only the final prediction but also the multi-granular interpretability for the decision-making process. Extensive experiments on the graph classification benchmarks with both synthetic and real-world datasets demonstrate the superiority of TIF in interpretability, while also delivering a competitive prediction performance akin to the state-of-the-art counterparts.

Problem

• Enhances interpretability of GNNs by revealing multi-granular decision-making subgraphs
• Addresses limitations of fixed-granularity methods in capturing varying graph relationships
• Proposes hierarchical tree framework for global and local graph interpretation

Innovation

• Hierarchical tree framework for multi-granular interpretability
• Graph coarsening and perturbation for diverse node representation
• Adaptive routing to identify informative decision paths