让深层图神经网络在异质图上更准，通过测试时混合浅层专家。
Mixture of Scope Experts at Test: Generalizing Deeper Graph Neural Networks with Shallow Variants
- 测试时动态选择浅层专家组合，适应不同节点的同质性差异。
- 在多个异质图数据集上提升深层GNN准确率，最高增益达12.3%。
- 适合追求高表达力又怕过拟合的图学习研究者与应用者。
异质图中相似节点倾向于不相连，这对图神经网络（GNN）构成挑战。增加GNN深度可扩大感受野，可能从高阶邻域发现同质性，但随着深度增加，模型性能会下降。尽管深层GNN具备更强表达力，其性能提升却仅比浅层变体微弱。通过理论与实证分析，我们系统揭示了随着深度增加，不同同质性水平节点的泛化偏好发生转变，导致深浅模型泛化模式不一致。基于此，我们提出测试时的范围专家混合（Moscat），在保持高表达力的同时显著改善深层GNN的泛化能力。实验表明，Moscat可灵活适配多种GNN架构，在广泛数据集上显著提升准确率。代码已开源：https://github.com/Hydrapse/moscat。
原文摘要 · Abstract (English)
Heterophilous graphs, where dissimilar nodes tend to connect, pose a challenge for graph neural networks (GNNs). Increasing the GNN depth can expand the scope (i.e., receptive field), potentially finding homophily from the higher-order neighborhoods. However, GNNs suffer from performance degradation as depth increases. Despite having better expressivity, state-of-the-art deeper GNNs achieve only marginal improvements compared to their shallow variants. Through theoretical and empirical analysis, we systematically demonstrate a shift in GNN generalization preferences across nodes with different homophily levels as depth increases. This creates a disparity in generalization patterns between GNN models with varying depth. Based on these findings, we propose to improve deeper GNN generalization while maintaining high expressivity by Mixture of scope experts at test (Moscat). Experimental results show that Moscat works flexibly with various GNNs across a wide range of datasets while significantly improving accuracy. Our code is available at (https://github.com/Hydrapse/moscat).
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。