Research map / Graph learning / Knowledge and temporal graphs
Hypergraph Foundation Model
Yue Gao, Yifan Feng, Shiquan Liu, Xiangmin Han, ShaoYi DU, Zongze Wu, Han Hu
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 |
|---|---|---|---|
| add neighborhood-prediction fine-tuning to the LM features (ablation) | Hyper-FM · Arxiv-Text | accuracy: 0.3231 -> 0.4189 | Improved |
| add neighborhood-prediction fine-tuning to the LM features (ablation) | Hyper-FM · Movielens-Text | accuracy: 0.2852 -> 0.3464 | Improved |
| add neighborhood-prediction fine-tuning to the LM features (ablation) | Hyper-FM · Pubmed-CC-Text | accuracy: 0.3967 -> 0.4009 | Improved |
| add neighborhood-prediction fine-tuning to the LM features (ablation) | Hyper-FM · 11 TAHG datasets | accuracy | Improved |
| use LM features instead of BoW (ablation) | Hyper-FM · 11 TAHG datasets | accuracy: 0.3142 -> 0.3297 on GoodBook-Text favors BoW | Mixed |
| use LM features instead of BoW (ablation) | Hyper-FM · Arxiv-Text | accuracy: 0.2786 -> 0.3231 | Improved |
| use multi-domain Hyper-FM pretraining instead of isolated sequential pretraining (IP) | HyperGCL + HGNN · Cora-CC-Text | accuracy: 0.3268 -> 0.3706 | Improved |
| use multi-domain Hyper-FM pretraining instead of isolated sequential pretraining (IP) | HyperGCL + HGNN · Arxiv-Text | accuracy: 0.3540 -> 0.4189 | Improved |
| isolated pretraining then fine-tuning instead of supervised HGNN only | HGNN · Cora-CC-Text | accuracy: 0.3570 -> 0.3268 | Hurt |
| Hyper-FM pretraining before fine-tuning vs supervised HGNN | HGNN · Pubmed-CC-Text | accuracy: 23.4% | Improved |
| Hyper-FM pretraining before fine-tuning vs supervised HGNN | HGNN · OGBN-Arxiv | accuracy: 14.3% | Improved |
| pretrain on Aminer-Text and transfer to PPI-Text (domain with different hyperedge degree distribution) | HGNN · PPI-Text | accuracy: 18.07% | Hurt |
Closest papers by meaning
- Beyond Graphs: Can Large Language Models Comprehend Hypergraphs? (ICLR 2025 · similarity 0.90)
- SPHINX: Structural Prediction using Hypergraph Inference Network (ICML 2025 · similarity 0.90)
- Training-Free Message Passing for Learning on Hypergraphs (ICLR 2025 · similarity 0.89)
- Hypergraph Representation Learning with Hyperlink Random Effects (NeurIPS 2026 · similarity 0.89)
- Hypergraph Neural Architecture Search (AAAI 2024 · similarity 0.89)
- Improved Expressivity of Hypergraph Neural Networks through High-Dimensional Generalized Weisfeiler-Leman Algorithms (ICML 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.
