无需分位数输入即可实现图神经网络的可靠不确定性量化。
Quantile-Free Uncertainty Quantification in Graph Neural Networks
- 采用双头架构分离预测与不确定性，通过无分位数联合损失训练。
- 在19个基准上平均覆盖率提升22%，区间宽度缩小50%。
- 适合需要高可靠性推理的图数据场景，如医疗或金融风控。
图神经网络（GNN）中的不确定性量化（UQ）在高风险领域至关重要，但仍是重大挑战。传统方法依赖消息传递中的可交换性假设，实际中常不成立，且可靠的UQ通常需昂贵的重采样或后处理校准。为此，我们提出无需分位数输入的预测区间图神经网络（QpiGNN），基于分位数回归（QR）直接优化覆盖概率和区间宽度。QpiGNN采用双头结构解耦预测与不确定性，仅用标签监督通过无分位数联合损失进行训练。该设计支持高效训练，理论保证渐近覆盖性和近似最优区间宽度。在19个合成与真实世界基准上的实验表明，相比基线，QpiGNN平均覆盖率达22%更高，区间宽度窄50%，且对噪声和结构变化具有鲁棒性。
原文摘要 · Abstract (English)
Uncertainty quantification (UQ) in graph neural networks (GNNs) is crucial in high-stakes domains but remains a significant challenge. In graph settings, message passing often relies on strong assumptions such as exchangeability, which are rarely satisfied in practice, and achieving reliable UQ typically requires costly resampling or post-hoc calibration. To address these issues, we introduce Quantile-free Prediction Interval GNN (QpiGNN), a framework that builds on quantile regression (QR) to enable GNN-based UQ by directly optimizing coverage and interval width without requiring quantile inputs or post-processing. QpiGNN employs a dual-head architecture that decouples prediction and uncertainty, and is trained with label-only supervision through a quantile-free joint loss. This design allows efficient training and yields robust prediction intervals, with theoretical guarantees of asymptotic coverage and near-optimal width under mild assumptions. Experiments on 19 synthetic and real-world benchmarks show QpiGNN achieves average 22% higher coverage and 50% narrower intervals than baselines, while ensuring efficiency and robustness to noise and structural shifts.
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。