[论文解读] TempME: Towards the Explainability of Temporal Graph Neural Networks via Motif Discovery

Jialin Chen, Rex Ying|arXiv (Cornell University)|Oct 30, 2023

Advanced Graph Neural Networks被引用 8

一句话总结

TempME 通过在信息瓶颈框架下发现连贯的时序模体来解释 TGNN，提升解释保真度，且可能提升预测性能。

ABSTRACT

Temporal graphs are widely used to model dynamic systems with time-varying interactions. In real-world scenarios, the underlying mechanisms of generating future interactions in dynamic systems are typically governed by a set of recurring substructures within the graph, known as temporal motifs. Despite the success and prevalence of current temporal graph neural networks (TGNN), it remains uncertain which temporal motifs are recognized as the significant indications that trigger a certain prediction from the model, which is a critical challenge for advancing the explainability and trustworthiness of current TGNNs. To address this challenge, we propose a novel approach, called Temporal Motifs Explainer (TempME), which uncovers the most pivotal temporal motifs guiding the prediction of TGNNs. Derived from the information bottleneck principle, TempME extracts the most interaction-related motifs while minimizing the amount of contained information to preserve the sparsity and succinctness of the explanation. Events in the explanations generated by TempME are verified to be more spatiotemporally correlated than those of existing approaches, providing more understandable insights. Extensive experiments validate the superiority of TempME, with up to 8.21% increase in terms of explanation accuracy across six real-world datasets and up to 22.96% increase in boosting the prediction Average Precision of current TGNNs.

研究动机与目标

• 说明时序图神经网络（TGNN）可解释性的必要性。
• 引入时序模体作为连贯、可解释的解释构建块。
• 开发基于 IB 的解释器 TempME，使用原假设模型来衡量模体显著性。
• 证明基于模体的解释能提高保真度并可能提升 TGNN 的预测性能。
• 提供一个可扩展的基于采样的框架来提取和编码时序模体。

提出的方法

• 在被查询的边周围取周边的回顾性时序模体样本。
• 使用事件匿名化、信息传递和时间感知池化的模体编码器对每个模体进行编码。
• 通过在解释准确性与压缩之间取得平衡的信息瓶颈目标来计算模体重要性分数。
• 引入一个空模型以比较模体分布并推导基于 KL 散度的正则化项。
• 对可微分训练使用 Concrete 松弛和使用 Bernoulli 掩码生成连贯的解释。
• 在六个真实世界数据集上对三种 TGNN 主干（TGAT、TGN、GraphMixer）进行解释评估。

实验结果

研究问题

• RQ1 哪些时序模体对 TGNN 对未来交互的预测影响最大？
• RQ2 基于模体的解释是否在不牺牲保真度的情况下具备连贯性与可人类理解性？
• RQ3 引入空模型是否能提高基于模体的解释的可靠性？
• RQ4 模体引导的解释是否在六个数据集上的 TGNN 体系结构中与提升的链路预测性能相关？

主要发现

• TempME 在六个真实世界数据集上实现解释准确性提升最高可达 8.21%。
• TempME 将 TGNN 的预测平均精度（AP）提升最高可达 22.96%。
• 时序模体相比基线方法能够提供更连贯、可解释的解释。
• TempME 在有无属性的数据集上均保持强解释性表现。
• 该方法在解释 TGN 时在所有六个数据集上达到最先进的 ACC-AUC 结果。
• 基于模体的解释在某些设定下也能提升 TGNN 的预测性能。