[论文解读] GISExplainer: On Explainability of Graph Neural Networks via Game-theoretic Interaction Subgraphs

Xingping Xian, Jing Liu+5 more|arXiv (Cornell University)|Sep 24, 2024

Scientific Computing and Data ManagementDecision Sciences被引用 3

一句话总结

本文提出 GraphGI，一种基于博弈论的图神经网络（GNN）解释方法，通过基于 Shapley 的博弈论交互值来引导迭代边选择，识别边之间交互强度最高的子图。该方法通过建模边级别的交互而非依赖于节点中心或独立特征归因，相较于最先进基线方法，在保真度和稀疏性方面表现更优。

ABSTRACT

Explainability is crucial for the application of black-box Graph Neural Networks (GNNs) in critical fields such as healthcare, finance, cybersecurity, and more. Various feature attribution methods, especially the perturbation-based methods, have been proposed to indicate how much each node/edge contributes to the model predictions. However, these methods fail to generate connected explanatory subgraphs that consider the causal interaction between edges within different coalition scales, which will result in unfaithful explanations. In our study, we propose GISExplainer, a novel game-theoretic interaction based explanation method that uncovers what the underlying GNNs have learned for node classification by discovering human-interpretable causal explanatory subgraphs. First, GISExplainer defines a causal attribution mechanism that considers the game-theoretic interaction of multi-granularity coalitions in candidate explanatory subgraph to quantify the causal effect of an edge on the prediction. Second, GISExplainer assumes that the coalitions with negative effects on the predictions are also significant for model interpretation, and the contribution of the computation graph stems from the combined influence of both positive and negative interactions within the coalitions. Then, GISExplainer regards the explanation task as a sequential decision process, in which a salient edges is successively selected and connected to the previously selected subgraph based on its causal effect to form an explanatory subgraph, ultimately striving for better explanations. Additionally, an efficiency optimization scheme is proposed for the causal attribution mechanism through coalition sampling. Extensive experiments demonstrate that GISExplainer achieves better performance than state-of-the-art approaches w.r.t. two quantitative metrics: Fidelity and Sparsity.

研究动机与目标

• 解决现有 GNN 解释方法将特征视为独立的局限性，忽略其相互作用。
• 通过识别捕捉边之间最强合作交互的子图，提升模型可解释性。
• 开发一种高效的博弈论交互值计算框架，使其可扩展至真实世界图结构。
• 通过在 GNN 的感受野内建模边级别交互，提升解释的保真度与稀疏性。

提出的方法

• GraphGI 将 GNN 解释建模为合作博弈，其中边作为参与者，其集体贡献通过博弈论交互值进行度量。
• 通过选择当前联盟中交互强度最大的边，逐步构建解释子图。
• 该方法使用 Shapley 值及其扩展形式计算交互强度，确保对合作效应的公平归因。
• 为降低计算成本，采用蒙特卡洛采样近似计算 Shapley 值与交互值。
• 解释子图被限制在预测目标的 L-跳邻域内，与 GNN 的消息传递机制保持一致。
• 该算法优先选择稀疏但高度交互的边集合，偏好单条有向边而非节点连接的子图。

实验结果

研究问题

• RQ1 能否有效建模 GNN 中的边级别交互，以提升解释保真度？
• RQ2 在 GNN 解释中，建模边之间的合作交互与独立特征归因相比有何优势？
• RQ3 博弈论交互值在多大程度上能增强 GNN 解释的稀疏性与可解释性？
• RQ4 高效的近似技术在保持高精度的同时，能否有效降低基于交互的解释的计算成本？

主要发现

• GraphGI 在合成数据集和真实世界数据集上均实现了最高的保真度与稀疏性，优于 SubgraphX、PGExplainer 等基线方法。
• 在 BA-shapes 与 Tree-cycles 数据集上，GraphGI 在 GIN 模型下实现了最佳保真度与稀疏性，表明其在不同图结构上的鲁棒性。
• 在 Cora 与 CiteSeer 等真实世界数据集中，GraphGI 即使在高稀疏度下仍保持高性能，表明其能有效识别关键交互子图。
• 由于其以边为中心、基于交互的筛选策略，GraphGI 的解释比 SubgraphX 和 PGExplainer 更为稀疏。
• 蒙特卡洛采样显著降低了计算时间，同时未牺牲解释质量，使方法可扩展至更大规模图结构。