Research map / Graph learning / GNN architectures
Hyperbolic Graph Neural Networks at Scale: A Meta Learning Approach
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 |
|---|---|---|---|
| Use H-GRAM instead of the strongest reported baseline | few-shot graph classification and link prediction · Synthetic Cycle/BA, ogbn-arxiv, Tissue-PPI, Fold-PPI, FirstMM-DB, Tree-of-Life | accuracy: .872 -> .883; .867 -> .873; .542 -> .555; .734 -> .746; .767 -> .779; .866 -> .888; .451 -> .472; .768 -> .786; .561 -> | Helped |
| Use H-GRAM rather than standard HNN models | node classification and link prediction on citation graphs · Cora, Pubmed, Citeseer | accuracy: .796 -> .827; .792 -> .790; .696 -> .716; .914 -> .896; .950 -> .924; .928 -> .936 | Mixed |
| Add meta-gradient updates (ablation) | H-GRAM · ogbn-arxiv, SG-DL | accuracy: .389 -> .472 | Helped |
| Add meta-gradient updates (ablation) | H-GRAM · Tissue-PPI, MG-SL | accuracy: .559 -> .786 | Helped |
| Add label prototypes (ablation) | H-GRAM · ogbn-arxiv, SG-DL | accuracy: .407 -> .472 | Helped |
| Add label prototypes (ablation) | H-GRAM · Tissue-PPI, MG-SL | accuracy: .762 -> .786 | Helped |
| Use HGCN instead of HMLP as the base encoder (ablation) | H-GRAM · Tissue-PPI, MG-SL | accuracy: .537 -> .786 | Helped |
| Use HGCN instead of HAT as the base encoder (ablation) | H-GRAM · Tissue-PPI, MG-SL | accuracy: .777 -> .786 | Helped |
| Increase few-shot examples during meta-testing | H-GRAM · Fold-PPI node classification and FirstMM-DB link prediction | accuracy | Helped |
| Increase neighborhood hops from K=2 to K=3 | H-GRAM · node classification on Fold-PPI | accuracy | Helped |
| Increase neighborhood hops from K=2 to K=3 | H-GRAM · link prediction on FirstMM-DB | accuracy | Hurt |
Closest papers by meaning
- Residual Hyperbolic Graph Convolution Networks (AAAI 2024 · similarity 0.90)
- Training-Free Message Passing for Learning on Hypergraphs (ICLR 2025 · similarity 0.90)
- Hgformer: Hyperbolic Graph Transformer for Collaborative Filtering (ICML 2025 · similarity 0.89)
- Improved Expressivity of Hypergraph Neural Networks through High-Dimensional Generalized Weisfeiler-Leman Algorithms (ICML 2025 · similarity 0.89)
- K-hop Hypergraph Neural Network: A Comprehensive Aggregation Approach (AAAI 2025 · similarity 0.89)
- THGNets: Constrained Temporal Hypergraphs and Graph Neural Networks in Hyperbolic Space for Information Diffusion Prediction (AAAI 2025 · 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.