Research map / Label-efficient and robust learning / Recognition and long-tail
SoftHGNN: Soft Hypergraph Neural Networks for General Visual Recognition
Mengqi Lei, Yihong Wu, Siqi Li, Xinhu Zheng, Juan Wang, Shaoyi Du, Yue Gao
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 SoftHGNN block at encoder end | PVT-Tiny · ImageNet-100 | Top-1: 82.7 -> 86.1 | Improved |
| add SoftHGNN block at encoder end | SwinT-Tiny · ImageNet-100 | Top-1: 88.0 -> 90.1 | Improved |
| add sparse hyperedge selection to SoftHGNN | PVT-Tiny · ImageNet-100 | Top-1: 86.1 -> 86.8 | Improved |
| add sparse hyperedge selection to SoftHGNN | SwinT-Tiny · ImageNet-100 | Top-1: 90.1 -> 90.9 | Improved |
| add k-NN HGNN block | PVT-Tiny · ImageNet-100 | Top-5: 95.4 -> 95.2 | Hurt |
| add self-attention block | PVT-Tiny · ImageNet-100 | Top-1: 82.7 -> 83.0 | No clear effect |
| add SoftHGNN block at encoder end | PVT-Tiny · ImageNet-1k | Top-1: 74.8 -> 75.6 | Improved |
| add sparse hyperedge selection to SoftHGNN | PVT-Tiny · ImageNet-1k | Top-1: 75.6 -> 75.9 | Improved |
| add SoftHGNN to multi-scale fusion | CCTrans Twins-PCPVT · ShanghaiTech Part-A | MAE: 54.8 -> 52.8 | Improved |
| add SoftHGNN to multi-scale fusion | CCTrans PVT · ShanghaiTech Part-B | MAE: 7.4 -> 6.8 | Improved |
| add SoftHGNN neck module | YOLO11-N · MS COCO | AP50:95: 38.2 -> 40.2 | Improved |
| add SoftHGNN neck module | YOLOv12-M · MS COCO | AP50:95: 51.9 -> 52.2 | Improved |
Closest papers by meaning
- Vision HGNN: An Image is More than a Graph of Nodes (ICCV 2023 · similarity 0.92)
- Hypergraph Vision Transformers: Images are More than Nodes, More than Edges (CVPR 2025 · similarity 0.88)
- DVHGNN: Multi-Scale Dilated Vision HGNN for Efficient Vision Recognition (CVPR 2025 · similarity 0.87)
- Deep Hypergraph Neural Networks with Tight Framelets (AAAI 2025 · similarity 0.86)
- LightHGNN: Distilling Hypergraph Neural Networks into MLPs for 100x Faster Inference (ICLR 2024 · similarity 0.86)
- HyperGLM: HyperGraph for Video Scene Graph Generation and Anticipation (CVPR 2025 · similarity 0.86)
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.
