Research map / Graph learning / GNN architectures
Topological 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 |
|---|---|---|---|
| Replace a GCN layer with TOGL | GCN · MNIST | accuracy: 90.0 -> 95.5 | Helped |
| Replace a GIN layer with TOGL | GIN · DD | accuracy: 70.8 -> 75.2 | Helped |
| Replace a GCN layer with TOGL | GCN · ENZYMES | accuracy: 65.8 -> 53.0 | Hurt |
| Replace a GIN layer with TOGL | GIN · MNIST | accuracy: 96.1 -> 96.1 | No clear effect |
| Replace a GCN layer with TOGL | GCN · CLUSTER | weighted accuracy: 57.0 -> 60.4 | Helped |
| Replace a GIN layer with TOGL | GIN · MOLHIV (structure-based) | ROC-AUC: 68.7 -> 65.1 | Hurt |
| Replace a GAT layer with TOGL | GAT · Pattern (structure-based) | accuracy: 73.1 -> 59.6 | Hurt |
| Use learnable rather than static filtrations (ablation) | GCN-1-TOGL-1 · IMDB-MULTI | accuracy: 48.3 \pm 4.9 -> 52.0 \pm 4.0 | Helped |
| Use learnable rather than static filtrations (ablation) | GCN-1-TOGL-1 · REDDIT-B (topology methods) | accuracy: 90.4 \pm 1.4 -> 90.1 \pm 0.8 | Hurt |
| Use an embedding function that models interactions between persistence-diagram points (ablation) | TOGL · Experiments across datasets | predictive performance | No clear effect |
| Add cycle and connected-component information (ablation) | TOGL on synthetic datasets · Cycles and Necklaces | accuracy | Helped |
| Use learned task-based filtrations rather than static degree-based filtration (ablation) | TOGL · Necklaces | accuracy | Helped |
Closest papers by meaning
- TopoTune: A Framework for Generalized Combinatorial Complex Neural Networks (ICML 2025 · similarity 0.89)
- TopoGCL: Topological Graph Contrastive Learning (AAAI 2024 · similarity 0.88)
- Boosting Graph Pooling with Persistent Homology (NeurIPS 2024 · similarity 0.88)
- Weisfeiler and Lehman Go Paths: Learning Topological Features via Path Complexes (AAAI 2024 · similarity 0.88)
- Graph Persistence goes Spectral (NeurIPS 2025 · similarity 0.88)
- Topological Blindspots: Understanding and Extending Topological Deep Learning Through the Lens of Expressivity (ICLR 2025 · similarity 0.88)
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.