Research map / Graph learning / GNN architectures
Manifold GCN: Diffusion-based Convolutional Neural Network for Manifold-valued Graphs
Martin Hanik, Gabriele Steidl, Christoph von Tycowicz, Alzheimer’s Disease Neuroimaging Initiative
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| use one-hot embedding instead of degree embedding | Manifold GCN · Synthetic graphs, 90 graphs | mean F1: 55.2 -> 67.8 | Improved |
| use one-hot embedding instead of degree embedding | Manifold GCN · Synthetic graphs, 2880 graphs | mean F1: 73.3 -> 85.2 | Improved |
| replace hyperbolic H100 with SPD(15) embedding space | Manifold GCN · Synthetic graphs, 6000 graphs | mean accuracy: 86.1% -> 85.0% | No clear effect |
| add manifold GCN (one-hot) versus H2H-GCN | H2H-GCN · Synthetic graphs, 6000 graphs | mean accuracy: 80.9% -> 86.1% | Improved |
| use manifold GCN instead of HGNN with little data | HGNN · Synthetic graphs, 90 graphs | mean F1: 41.8 -> 67.8 | Improved |
| add tMLPs (instead of manifold FCL) (ablation) | Manifold GCN · ADNI normals | accuracy: 74.5 -> 75.3 | Improved |
| add diffusion layers (ablation) | Manifold GCN · ADNI normals | accuracy: 70.5 -> 75.3 | Improved |
| add node-wise MLPs (ablation) | Manifold GCN · ADNI normals | accuracy: 70.5 -> 75.3 | Improved |
| add equivariance to tMLP (ablation) | Manifold GCN · ADNI normals | accuracy: 70.8 -> 75.3 | Improved |
| use sphere-aware S2 normals instead of Euclidean GCN on normals | Normals features · ADNI | accuracy: 70.9 -> 75.3 | Improved |
| use differential coordinates instead of normals as features | Manifold GCN · ADNI | accuracy: 75.3 -> 76.9 | Improved |
| use manifold GCN instead of DiffusionNet on same differential coordinates | Differential Coords. · ADNI | accuracy: 67.2 -> 76.9 | Improved |
Closest papers by meaning
- Learning to Approximate Adaptive Kernel Convolution on Graphs (AAAI 2024 · similarity 0.89)
- A Manifold Perspective on the Statistical Generalization of Graph Neural Networks (ICML 2025 · similarity 0.88)
- Building Transformation Layers for Riemannian Neural Networks (NeurIPS 2026 · similarity 0.88)
- Toward a Unified Geometry Understanding : Riemannian Diffusion Framework for Graph Generation and Prediction (NeurIPS 2025 · similarity 0.88)
- Bundle Neural Network for message diffusion on graphs (ICLR 2025 · similarity 0.87)
- Consistent Geometric Deep Learning via Hilbert Bundles and Cellular Sheaves (NeurIPS 2026 · similarity 0.87)
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.
