arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2401.03350cs.LGstat.ML
收藏

面向图神经网络的认知不确定性精准可扩展估计

Accurate and Scalable Estimation of Epistemic Uncertainty for Graph Neural Networks

University of Michigan(密歇根大学);Lawrence Livermore National Laboratory(劳伦斯利弗莫尔国家实验室);Amazon(亚马逊公司)

Puja Trivedi, Mark Heimann, Rushil Anirudh, Danai Koutra, Jayaraman J. Thiagarajan

AI总结:

针对图神经网络在分布偏移下不确定性估计可靠性不足的问题,提出基于图锚定策略的G-$Δ$UQ训练框架,无需全随机网络即可在预训练模型上应用,显著提升多类偏移场景下的校准及相关不确定性任务性能。

摘要:

尽管图神经网络(GNNs)已被广泛应用于节点和图表示学习任务,但分布偏移下GNN不确定性估计的可靠性仍未得到充分研究。事实上,虽然事后校准策略可用于提升分布内校准效果,但它们未必能改善分布偏移下的校准表现。而能让GNN生成更优内在不确定性估计的技术价值尤为突出,因为这类技术后续始终可与事后策略结合使用。因此,本研究提出G-$Δ$UQ——一种旨在提升GNN内在不确定性估计能力的新型训练框架。该框架通过全新的图锚定策略,将随机数据中心化原理适配到图数据场景,且能够支持部分随机GNN。尽管主流观点认为,要获得可靠的估计结果必须使用全随机网络,但我们发现,我们的锚定策略在采样假设时诱导的功能多样性使得全随机网络不再必要,还让我们能够在预训练模型上应用G-$Δ$UQ。我们在协变量偏移、概念偏移和图规模偏移场景下开展了大量评估,结果表明G-$Δ$UQ能提升节点分类和图分类任务中GNN的校准效果。此外,它还提升了分布外检测和泛化间隙估计这类基于不确定性的任务的性能。总体而言,本研究为GNN的不确定性估计提供了洞见,并证明了G-$Δ$UQ在获取可靠估计结果方面的实用性。

英文摘要:

While graph neural networks (GNNs) are widely used for node and graph representation learning tasks, the reliability of GNN uncertainty estimates under distribution shifts remains relatively under-explored. Indeed, while post-hoc calibration strategies can be used to improve in-distribution calibration, they need not also improve calibration under distribution shift. However, techniques which produce GNNs with better intrinsic uncertainty estimates are particularly valuable, as they can always be combined with post-hoc strategies later. Therefore, in this work, we propose G-$Δ$UQ, a novel training framework designed to improve intrinsic GNN uncertainty estimates. Our framework adapts the principle of stochastic data centering to graph data through novel graph anchoring strategies, and is able to support partially stochastic GNNs. While, the prevalent wisdom is that fully stochastic networks are necessary to obtain reliable estimates, we find that the functional diversity induced by our anchoring strategies when sampling hypotheses renders this unnecessary and allows us to support G-$Δ$UQ on pretrained models. Indeed, through extensive evaluation under covariate, concept and graph size shifts, we show that G-$Δ$UQ leads to better calibrated GNNs for node and graph classification. Further, it also improves performance on the uncertainty-based tasks of out-of-distribution detection and generalization gap estimation. Overall, our work provides insights into uncertainty estimation for GNNs, and demonstrates the utility of G-$Δ$UQ in obtaining reliable estimates.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑