Mitigating Over-squashing without Rewiring: A Sheaf Effective Resistance Perspective

arXiv:2610.04157 · stat.ML, cs.LG · Submitted 2026-10-03 · Read on arXiv

stat.ML, cs.LG

Submitted: 2026-10-03

Updated: 2026-10-03

Comments: Accepted to NeurIPS 2026

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Over-squashing — 1× in this paper · explained in 3 episodes

Sources

• A Note on Over-Smoothing for Graph Neural Networks
• Can You Hear Me Now? A Benchmark for Long-Range Graph Propagation
• Short-Range Oversquashing
• TUDataset: A collection of benchmark datasets for learning with graphs
• Geom-GCN: Geometric Graph Convolutional Networks
• A critical look at the evaluation of GNNs under heterophily: Are we really making progress?
• Masked Label Prediction: Unified Message Passing Model for Semi-Supervised Classification
• Graph Attention Networks

Related papers

• Behavior of prediction performance metrics with rare events
• Optimal Estimation of Generic Dynamics by Path-Dependent Neural Jump ODEs
• A Posterior-Dynamics Framework for Imaging Inverse Problems with Pretrained Diffusion Priors
• Efficient quadratic entropy with distance sketches
• Embedding-Bias in Conditional Independence Testing
• From Log-Odds to Shapley Values: An Explanatory Geometry for the Weighted Naive Bayes Classifier