Research map / Graph learning / GNN architectures
Hierarchical Graph Capsule Network
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 disentangled latent-factor graph capsules (ablation) | HGCN · MUTAG | accuracy: 91.46$\pm$5.77 -> 93.16$\pm$6.10 | Helped |
| Add the residual connection between capsule layers (ablation) | HGCN · MUTAG | accuracy: 92.08$\pm$5.10 -> 93.16$\pm$6.10 | Helped |
| Add auxiliary graph reconstruction (ablation) | HGCN · MUTAG | accuracy: 92.11$\pm$7.13 -> 93.16$\pm$6.10 | Helped |
| Add the residual connection between capsule layers (ablation) | HGCN · NCI1 | accuracy: 85.28$\pm$1.37 -> 84.87$\pm$1.68 | Hurt |
| Add auxiliary graph reconstruction (ablation) | HGCN · COLLAB | accuracy: 83.08$\pm$1.69 -> 82.86$\pm$1.81 | Hurt |
| Use K=8 instead of K=4 latent factors (ablation) | HGCN · COLLAB | accuracy: 82.86$\pm$1.81 -> 83.10$\pm$1.80 | Mixed |
| Use R=4 instead of R=3 routing iterations (ablation) | HGCN · D&D | accuracy: 80.99$\pm$2.58 -> 81.15$\pm$3.79 | Helped |
| Use HGCN instead of the strongest listed baseline | graph classification · ENZYMES | accuracy: 70.17$\pm$5.58 -> 78.00$\pm$4.89 | Helped |
Closest papers by meaning
- Capsule Graph Neural Network (ICLR 2019 · similarity 0.91)
- H$^3$GNNs: Harmonizing Heterophily and Homophily in GNNs via Self-Supervised Node Encoding (ICLR 2026 · similarity 0.87)
- A New Perspective on "How Graph Neural Networks Go Beyond Weisfeiler-Lehman?" (ICLR 2022 · similarity 0.87)
- Heterogeneous Graph Structure Learning for Graph Neural Networks (AAAI 2021 · similarity 0.87)
- Improving Graph Neural Networks by Learning Continuous Edge Directions (ICLR 2025 · similarity 0.86)
- Exploring Consistency in Graph Representations: from Graph Kernels to Graph Neural Networks (NeurIPS 2024 · similarity 0.85)
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.