Research map / Graph learning / GNN architectures
Revisiting Heterophily For Graph Neural Networks
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 |
|---|---|---|---|
| add high-pass and identity channels with adaptive mixing to low-pass-only GCN (ablation) | ACM-GCN · Squirrel | accuracy: 44.76 -> 58.02 | Helped |
| add high-pass and identity channels with adaptive mixing to low-pass-only GCN (ablation) | ACM-GCN · Texas | accuracy: 83.11 -> 95.08 | Helped |
| add high-pass and identity channels with adaptive mixing to low-pass-only SGC-1 (ablation) | ACM-SGC-1 · Cornell | accuracy: 70.98 -> 93.77 | Helped |
| add high-pass and identity channels with mixing to GCN (ablation) | ACM-GCN · Cora | accuracy: 87.78 -> 88.95 | Helped |
| add adaptive mixing over summing the three channels (ablation) | ACM-GCN · Texas | accuracy: 95.25 -> 95.08 | Hurt |
| add adaptive mixing over summing the three channels (ablation) | ACM-GCN · Squirrel | accuracy: 54.78 -> 58.02 | Helped |
| add adaptive mixing over summing the three channels (ablation) | ACM-GCN · Cornell | accuracy: 91.64 -> 94.75 | Helped |
| add identity channel to LP+HP with mixing (ablation) | ACM-GCN · Squirrel | accuracy: 57.2 -> 58.02 | Helped |
| add identity channel to LP+HP with mixing (ablation) | ACM-GCN · Texas | accuracy: 89.19 -> 95.08 | Helped |
| add high-pass channel to LP+identity with mixing (ablation) | ACM-GCN · Squirrel | accuracy: 53.97 -> 58.02 | Helped |
| add high-pass channel to LP+identity with mixing (ablation) | ACM-GCN · Texas | accuracy: 94.1 -> 95.08 | No clear effect |
| add ACM components to GCN (ablation) | ACM-GCN · runtime per epoch | accuracy: 3.67 -> 8.25 ms | Hurt |
Closest papers by meaning
- On the Impact of Feature Heterophily on Link Prediction with Graph Neural Networks (NeurIPS 2024 · similarity 0.93)
- Understanding Heterophily for Graph Neural Networks (ICML 2024 · similarity 0.92)
- What Is Missing For Graph Homophily? Disentangling Graph Homophily For Graph Neural Networks (NeurIPS 2024 · similarity 0.92)
- Making Classic GNNs Strong Baselines Across Varying Homophily: A Smoothness–Generalization Perspective (NeurIPS 2025 · similarity 0.91)
- Integrating Co-Training with Edge Discrimination to Enhance Graph Neural Networks Under Heterophily (AAAI 2025 · similarity 0.91)
- When Hypergraph Meets Heterophily: New Benchmark Datasets and Baseline (AAAI 2025 · similarity 0.90)
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.