AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

Generalizing Graph Neural Networks on Out-of-Distribution Graphs

Shaohua Fan, Xiao Wang, Chuan Shi, Peng Cui, Bai Wang

TPAMI 2023 · Graph learning › Graph self-supervised learning · approach: Graph out-of-distribution generalization · 136 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
add StableGNN framework (high-level variable learning + CVD) to GraphSAGEGraphSAGE · Synthetic house-motif (mu=0.9)accuracy: 65.2 -> 70.35Improved
add StableGNN framework to GCNGCN · Synthetic house-motif (mu=0.9)accuracy: 69.68 -> 76.35Improved
increase spurious correlation degree mu from 0.6 to 0.9GraphSAGE · Synthetic house-motifF1: 67.83 -> 54.24Hurt
add StableGNN to GraphSAGEGraphSAGE · MolhivROC-AUC: 75.78 -> 77.63Improved
add StableGNN to GraphSAGEGraphSAGE · MolesolRMSE: 1.1852 -> 1.0092Improved
add StableGNN to GCNGCN · MolesolRMSE: 1.153 -> 0.9638Improved
add StableGNN to GraphSAGEGraphSAGE · Moltox21ROC-AUC: 68.88 -> 69.14No clear effect
add CVD regularizer to StableGNN-NoCVD (ablation)StableGNN-NoCVD · MolhivROC-AUC: 76.47 -> 77.63Improved
add CVD regularizer to StableGNN-NoCVD (ablation)StableGNN-NoCVD · MolesolRMSE: 1.0819 -> 1.022Improved
use cluster-level CVD instead of per-dimension decorrelation (ablation)StableGNN-NoCVD · MolhivROC-AUC: 75.32 -> 77.63Improved
use sample reweighting CVD instead of HSIC disentanglement (ablation)StableGNN-NoCVD · MolbbbpROC-AUC: 64.21 -> 68.47Improved
add per-dimension decorrelation to GraphSAGE (ablation)GraphSAGE · MolbaceROC-AUC: 78.51 -> 77.34Hurt

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