CUQ-GNN: Committee-based Graph Uncertainty Quantification using Posterior Networks

TL;DR

This paper introduces CUQ-GNN, a novel graph neural network model that combines committee-based methods with posterior networks to improve uncertainty quantification in node classification tasks, addressing limitations of previous approaches.

Contribution

The paper proposes CUQ-GNN, a new framework that enhances uncertainty estimation in graph neural networks by integrating committee-based methods with posterior networks, providing more flexible and domain-adaptive uncertainty measures.

Findings

CUQ-GNN outperforms GPN and other methods on benchmark datasets.

It produces more reliable and meaningful uncertainty estimates.

The approach is adaptable to domain-specific uncertainty requirements.

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 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 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.