Abstract
In this work, we study the influence of domain-specific characteristics when defining a meaningful notion of predictive uncertainty on graph data. Previously, the so-called Graph Posterior Network (GPN) model has been proposed to quantify uncertainty in node classification tasks. Given a graph, it uses Normalizing Flows (NFs) to estimate class densities for each node independently and converts those densities into Dirichlet pseudo-counts, which are then dispersed through the graph using the personalized Page-Rank (PPR) algorithm. The architecture of GPNs is motivated by a set of three axioms on the properties of its uncertainty estimates. We show that those axioms are not always satisfied in practice and therefore propose the family of Committe-based Uncertainty Quantification Graph Neural Networks (CUQ-GNNs), which combine standard Graph Neural Networks (GNNs) with the NF-based uncertainty estimation of Posterior Networks (PostNets). This approach adapts more flexibly to domain-specific demands on the properties of uncertainty estimates. We compare CUQ-GNN against GPN and other uncertainty quantification approaches on common node classification benchmarks and show that it is effective at producing useful uncertainty estimates.
Dokumententyp: | Konferenzbeitrag (Paper) |
---|---|
Keywords: | uncertainty quantification, graph neural networks |
Fakultät: | Mathematik, Informatik und Statistik > Informatik > Künstliche Intelligenz und Maschinelles Lernen |
Themengebiete: | 000 Informatik, Informationswissenschaft, allgemeine Werke > 004 Informatik |
URN: | urn:nbn:de:bvb:19-epub-121702-6 |
ISBN: | 978-3-031-70370-6 |
ISSN: | 0302-9743 |
Ort: | Cham |
Sprache: | Englisch |
Dokumenten ID: | 121702 |
Datum der Veröffentlichung auf Open Access LMU: | 09. Okt. 2024 07:37 |
Letzte Änderungen: | 20. Nov. 2024 15:44 |