AI Paper Atlas

Research map / Graph learning / GNN architectures

Uncertainty Quantification over Graph with Conformalized Graph Neural Networks

Kexin Huang, Ying Jin, Emmanuel Candes, Jure Leskovec

NeurIPS 2023 · Graph learning › GNN architectures · approach: GNN generalization theory · 16 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace inefficiency loss with standard prediction loss in correction step (ablation)CF-GNN · Cora (GCN)prediction set size: 1.76 -> 2.42 when removedImproved
replace inefficiency loss with standard prediction loss (ablation)CF-GNN · Anaheim (GCN)interval length: 2.17 -> 2.23 when removedImproved
use GNN instead of MLP as correction model (ablation)CF-GNN · Cora (GCN)prediction set size: 2.35 -> 1.76Improved
use GNN instead of MLP as correction model (ablation)CF-GNN · Anaheim (GCN)interval length: 2.32 -> 2.17Improved
add topology-aware correction and inefficiency loss to plain CPCF-GNN · Cora (GCN)prediction set size: 3.80 -> 1.76Improved
add CF-GNN correction to plain CP for regressionCF-GNN · Election / Chicago (GCN)interval length: 0.90 -> 0.90; 2.05 -> 2.04No clear effect
CF-GNN vs plain CP regression on AnaheimCF-GNN · Anaheim (GCN)interval length: 2.89 -> 2.17Improved
CF-GNN vs plain CP on conditional coverage over PageRank slices (classification)CF-GNN · Coraworst-slice coverage: 0.78 -> 0.94Improved
CF-GNN vs plain CP conditional coverage for regressionCF-GNN · Twitchworst-slice coverage: 0.94 -> 0.94 (PageRank)No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy