Research map / Graph learning / GNN architectures
Hyper-SAGNN: a self-attention based graph neural network for hypergraphs
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 Hyper-SAGNN-W instead of DHNE | Hyperedge network reconstruction · GPS | AUC: 0.959 -> 0.976 | Helped |
| Use Hyper-SAGNN-W instead of DHNE | Hyperedge network reconstruction · MovieLens | AUPR: 0.878 -> 0.986 | Helped |
| Use Hyper-SAGNN-W instead of DHNE | Hyperedge network reconstruction · drug | AUPR: 0.873 -> 0.945 | Helped |
| Use Hyper-SAGNN-W instead of DHNE | Hyperedge network reconstruction · wordnet | AUPR: 0.953 -> 0.956 | Helped |
| Train with a mixture of edges and hyperedges (ablation) | Hyper-SAGNN-E · GPS, MovieLens, drug, wordnet hyperedge prediction | AUC and AUPR | No clear effect |
| Replace average pooling with min pooling and fine-tune (ablation) | Hyper-SAGNN · drug outsider identification | top-1 accuracy: 78.5% to 81.9% | Helped |
| Replace average pooling with min pooling and fine-tune (ablation) | Hyper-SAGNN · drug hyperedge prediction | AUC: 0.955 to 0.935 | Hurt |
| Include the self-attention term αii (ablation) | Hyper-SAGNN · Multiple datasets; hyperedge prediction and node classification | performance | No clear effect |
| Use dynamic embeddings directly for probability scoring instead of static-dynamic differences (ablation) | Hyper-SAGNN variant type II · Multiple datasets; reconstruction, link prediction, and node classification | performance | Hurt |
Closest papers by meaning
- HypeBoy: Generative Self-Supervised Representation Learning on Hypergraphs (ICLR 2024 · similarity 0.90)
- Hypergraph Neural Architecture Search (AAAI 2024 · similarity 0.90)
- When Hypergraph Meets Heterophily: New Benchmark Datasets and Baseline (AAAI 2025 · similarity 0.89)
- K-hop Hypergraph Neural Network: A Comprehensive Aggregation Approach (AAAI 2025 · similarity 0.89)
- Deep Hypergraph Neural Networks with Tight Framelets (AAAI 2025 · similarity 0.88)
- Training-Free Message Passing for Learning on Hypergraphs (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.