Research map / Graph learning / GNN architectures
Universal Graph Convolutional 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 2-hop network branch to U-GCN without it (ablation) | U-GCN-3 · average over 8 datasets | accuracy (%): 65.40 -> 67.89 | Helped |
| add kNN network branch to GCN (1-hop) (ablation) | GCN · average over 8 datasets | accuracy (%): 58.96 -> 65.40 | Helped |
| use 2-hop neighbors instead of 1-hop in GCN (ablation) | GCN · average over 8 datasets | accuracy (%): 58.96 -> 62.33 | Helped |
| use kNN neighbors instead of 1-hop in GCN (ablation) | GCN · average over 8 datasets | accuracy (%): 58.96 -> 62.08 | Helped |
| use 2-hop neighbors instead of 1-hop in GCN (ablation) | GCN · Cora | accuracy (%): 82.93 -> 74.40 | Hurt |
| use 2-hop neighbors instead of 1-hop in GCN (ablation) | GCN · Cornell | accuracy (%): 46.51 -> 56.59 | Helped |
| use kNN neighbors instead of 1-hop in GCN (ablation) | GCN · Cornell | accuracy (%): 46.51 -> 70.54 | Helped |
| use kNN neighbors instead of 1-hop in GCN (ablation) | GCN · Chameleon | accuracy (%): 52.32 -> 38.60 | Hurt |
| add 2-hop branch to U-GCN-3 (ablation) | U-GCN-3 · Texas | accuracy (%): 62.79 -> 71.72 | Helped |
| add 2-hop branch to U-GCN-3 (ablation) | U-GCN-3 · Cornell | accuracy (%): 65.12 -> 69.77 | Helped |
| add 2-hop branch to U-GCN-3 (ablation) | U-GCN-3 · Cora | accuracy (%): 83.67 -> 84.00 | Helped |
| add kNN branch to U-GCN-1 style multi-type setup (U-GCN-2 vs U-GCN) (ablation) | U-GCN-2 · average over 8 datasets | accuracy (%): 62.08 -> 67.89 | Helped |
Closest papers by meaning
- Powerful Graph Convolutional Networks with Adaptive Propagation Mechanism for Homophily and Heterophily (AAAI 2022 · similarity 0.94)
- Block Modeling-Guided Graph Convolutional Neural Networks (AAAI 2022 · similarity 0.91)
- Deeper Insights Into Graph Convolutional Networks for Semi-Supervised Learning (AAAI 2018 · similarity 0.91)
- L2-GCN: Layer-Wise and Learned Efficient Training of Graph Convolutional Networks (CVPR 2020 · similarity 0.90)
- Making Classic GNNs Strong Baselines Across Varying Homophily: A Smoothness–Generalization Perspective (NeurIPS 2025 · similarity 0.89)
- Refining Latent Homophilic Structures over Heterophilic Graphs for Robust Graph Convolution Networks (AAAI 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.