AI Paper Atlas

Research map / Graph learning / GNN architectures

Half-Hop: A graph upsampling approach for slowing down message passing

Mehdi Azabou, Venkataramana Ganesh, Shantanu Thakoor, Chi-Heng Lin, Lakshmi Sathidevi, Ran Liu, Michal Valko, Petar Veličković, Eva L Dyer

ICML 2023 · Graph learning › GNN architectures · approach: Graph rewiring and scalable message passing · 0 citations (OpenAlex)

Read the paper

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 Half-Hop to GCNGCN · Texas, Wisconsin, Actor, Squirrel, Chameleon, Cornellaccuracy: +16.75, +19.04, +7.80, +15.67, +21.80, +2.70Improved
Add Half-Hop to GATGAT · Texas, Wisconsin, Actor, Squirrel, Chameleon, Cornellaccuracy: +28.38, +34.12, +9.26, +9.58, +12.75, +10.81Improved
Add Half-Hop to GraphSAGEGraphSAGE · Texas, Wisconsin, Actor, Squirrel, Chameleon, Cornellaccuracy: +3.51, +4.70, +2.59, +3.64, +4.25, -1.35Mixed
Add Half-Hop to GCNGCN · Amazon Computers, Amazon Photos, Coauthor CS, WikiCSaccuracy: +0.70, +0.93, +0.65, +0.64Improved
Add Half-Hop to GraphSAGEGraphSAGE · Amazon Computers, Amazon Photos, Coauthor CS, WikiCSaccuracy: +1.81, -0.48, +0.02, -0.86Mixed
Use Half-Hop as an additional augmentation alongside feature and edge dropoutBGRL · Amazon Computers, Amazon Photos, Coauthor CS, Coauthor Physics, WikiCSaccuracy: 91.02 vs 90.34; 93.88 vs 93.17; 93.61 vs 93.31; 95.75 vs 95.73; 80.76 vs 79.98Improved
Use Half-Hop as an additional augmentation alongside feature and edge dropoutGRACE · Amazon Computers, Amazon Photos, Coauthor CS, WikiCSaccuracy: 91.11 vs 89.53; 94.21 vs 92.78; 93.59 vs 91.12; 80.77 vs 80.14Improved
Train with Half-Hop and evaluate on the Half-Hopped graphHH-GRACE · Amazon Photos, Coauthor CSaccuracy: 94.89 vs 92.78; 94.76 vs 91.12Improved
Remove the backward edge from target to slow node (ablation)GCN + Half-Hop · Texas, Actor, Cornellaccuracy: 71.71 -> 68.8; 33.35 -> 32.17; 63.42 -> 57.66Hurt
Add a slow-node-to-source edge, allowing reverse path connectivity (ablation)GCN + Half-Hop · Texas, Actor, Cornellaccuracy: 71.71 -> 58.47; 33.35 -> 31.93; 63.42 -> 42.16Hurt
Replace interpolated slow-node features with zeros (ablation)GCN + Half-Hop · Texas, Actor, Cornellaccuracy: 72.88 -> 61.80; 33.39 -> 28.93; 63.33 -> 49.55Hurt
Replace interpolated slow-node features with uniform random features (ablation)GCN + Half-Hop · Texas, Actor, Cornellaccuracy: 72.88 -> 53.33; 33.39 -> 24.70; 63.33 -> 37.48Hurt

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