Research map / Graph learning / GNN architectures
Large Scale Learning on Non-Homophilous Graphs: New Benchmarks and Strong Simple Methods
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| separately embed adjacency and features then mix (LINKX) instead of concatenating A and X into one MLP (ablation) | LINKX · arXiv-year | test accuracy: 54.15 -> 56.00 | Helped |
| separately embed adjacency and features then mix (LINKX) instead of concatenation (ablation) | LINKX · snap-patents | test accuracy: 59.12 -> 61.95 | Helped |
| separately embed adjacency and features then mix (LINKX) instead of concatenation (ablation) | LINKX · genius | test ROC AUC: 91.61 -> 90.77 | Hurt |
| separately embed adjacency and features then mix (LINKX) instead of concatenation (ablation) | LINKX · Penn94 | test accuracy: 84.64 -> 84.71 | No clear effect |
| add node features to LINK via LINKX (combine adjacency and feature embeddings) | LINK · genius | test ROC AUC: 73.56 -> 90.77 | Helped |
| train GCNJK/MixHop with Cluster-GCN or GraphSAINT minibatching instead of full batch | MixHop · snap-patents | test accuracy: 52.16 -> 34.21 | Hurt |
| train LINKX with simple i.i.d. node minibatching instead of full batch | LINKX · Penn94 | test accuracy: 84.71 -> 84.50 | No clear effect |
| use two-hop instead of one-hop propagation for C&S | C&S · pokec | test accuracy: 62.35 -> 81.69 | Helped |
Closest papers by meaning
- Graph homophily booster: Reimagining the role of discrete features in heterophilic graph learning (ICLR 2026 · similarity 0.91)
- Is Homophily a Necessity for Graph Neural Networks? (ICLR 2022 · similarity 0.91)
- What Is Missing For Graph Homophily? Disentangling Graph Homophily For Graph Neural Networks (NeurIPS 2024 · similarity 0.90)
- Making Classic GNNs Strong Baselines Across Varying Homophily: A Smoothness–Generalization Perspective (NeurIPS 2025 · similarity 0.90)
- When Hypergraph Meets Heterophily: New Benchmark Datasets and Baseline (AAAI 2025 · similarity 0.90)
- On the Impact of Feature Heterophily on Link Prediction with Graph Neural Networks (NeurIPS 2024 · similarity 0.89)
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.