AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

Improving Subgraph Recognition With Variational Graph Information Bottleneck

Junchi Yu, Jie Cao, Ran He

CVPR 2022 · Graph learning › Graph self-supervised learning · approach: Semi-supervised node representation learning · 62 citations (OpenAlex)

Check 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 GIB mutual-information estimation with VGIB noise injection and its variational objectiveGCN subgraph recognition · QED, HLM-CLint, MLM-CLint, and RLM-CLintabsolute property divergence: QED: 0.38±0.12 -> 0.32±0.12; HLM-CLint: 0.37±0.30 -> 0.34±0.28; MLM-CLint: 0.72±0.55 -> 0.69±0.58; RLM-CLint: 1.15±0.68 Improved
Use VGIB instead of GIBGCN subgraph recognition · QEDtraining time: 1712.93s -> 146.53sImproved
Use VGIB to generate explanations instead of the strongest listed baselineGCN explanations · RLM-CLintFidelity+: 0.694 -> 0.765Improved
Use VGIB to generate explanations instead of the strongest listed baselineGCN explanations · HLM-CLintFidelity+: 0.782 -> 0.792Improved
Use VGIB to generate explanations instead of the strongest listed baselineGCN explanations · QEDFidelity+: 0.602 -> 0.627Improved
Use VGIB to generate explanations instead of the strongest listed baselineGCN explanations · DRD2Fidelity+: 0.74 -> 0.756Improved
Use VGIB rather than the strongest listed baselineGCN explanations · RLM-CLintFidelity-: 0.478 -> 0.463Improved
Use VGIB rather than the strongest listed baselineGCN explanations · HLM-CLintFidelity-: 0.616 -> 0.579Improved
Use VGIB rather than the strongest listed baselineGCN explanations · QEDFidelity-: 0.498 -> 0.487Improved
Use VGIB rather than the strongest listed baselineGCN explanations · DRD2Fidelity-: 0.428 -> 0.424Improved
Add VGIB subgraph recognition to GINGIN with sum pooling · MUTAGaccuracy: 0.84±0.06 -> 0.86±0.08Improved
Add VGIB subgraph recognition to GINGIN with sum pooling · PROTEINSaccuracy: 0.74±0.05 -> 0.75±0.04Improved

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