Research map / Graph learning / Graph self-supervised learning
HyperGCN: A New Method For Training Graph Convolutional Networks on 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 |
|---|---|---|---|
| replace clique expansion (HGNN) with HyperGCN mediator expansion | HyperGCN · DBLP co-authorship SSL | test error (%): 25.65 -> 24.09 | Helped |
| replace clique expansion (HGNN) with HyperGCN mediator expansion | HyperGCN · Pubmed co-citation SSL | test error (%): 29.41 -> 25.56 | Helped |
| replace clique expansion (HGNN) with HyperGCN | HyperGCN · Cora co-citation SSL | test error (%): 32.41 -> 32.37 | No clear effect |
| add mediators to 1-HyperGCN (ablation) | HyperGCN · DBLP co-authorship SSL | test error (%): 33.87 -> 24.09 | Helped |
| recompute Laplacian each epoch from learned hidden features instead of initial features only (ablation) | HyperGCN vs FastHyperGCN · DBLP co-authorship SSL | test error (%): 27.34 -> 24.09 | Helped |
| use FastHyperGCN instead of HGNN | FastHyperGCN · DBLP | epoch training time (s): 0.115s -> 0.035s | Helped |
| replace HGNN with HyperGCN at low noise (eta=0.75) | HyperGCN · Synthetic noisy hyperedges | test error (%): 15.92 -> 22.44 | Hurt |
| replace HGNN with HyperGCN at high noise (eta=0.50) | HyperGCN · Synthetic noisy hyperedges | test error (%): 44.25 -> 37.89 | Helped |
| replace HGNN with HyperGCN on DBLP subset with large hyperedges | HyperGCN · sDBLP | test error (%): 45.27 -> 41.64 | Helped |
| add hypergraph structure via HyperGCN to MLP | MLP · DBLP co-authorship SSL | test error (%): 37.77 -> 24.09 | Helped |
| replace HGNN with HyperGCN | HyperGCN · Densest k-subhypergraph Cora co-authorship | density: 437 -> 504 | Helped |
Closest papers by meaning
- DHG-Bench: A Comprehensive Benchmark for Deep Hypergraph Learning (ICLR 2026 · similarity 0.90)
- Hypergraph-enhanced Dual Semi-supervised Graph Classification (ICML 2024 · similarity 0.90)
- K-hop Hypergraph Neural Network: A Comprehensive Aggregation Approach (AAAI 2025 · similarity 0.89)
- Deep Hypergraph Neural Networks with Tight Framelets (AAAI 2025 · similarity 0.89)
- When Hypergraph Meets Heterophily: New Benchmark Datasets and Baseline (AAAI 2025 · similarity 0.89)
- 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.