Research map / Label-efficient and robust learning / Recognition and long-tail
Hyperbolic Representation Learning: Revisiting and Advancing
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 |
|---|---|---|---|
| Add HIE to HGCN | HGCN · Disease node classification | F1-score: 90.6±1.2 -> 95.0±0.8 | Helped |
| Add HIE to HGCN | HGCN · Airport node classification | F1-score: 94.0±0.6 -> 94.1±0.8 | Helped |
| Add HIE to HGCN | HGCN · Citeseer node classification | accuracy: 67.6±0.4 -> 74.1±0.3 | Helped |
| Add HIE to HGCN | HGCN · Cora node classification | accuracy: 78.5±0.6 -> 83.0±0.3 | Helped |
| Add HIE to the hyperbolic shallow model | Hyperbolic shallow model · Disease link prediction, 25% training links, dimension 256 | AUC: 55.0±0.4 -> 66.8±0.4 | Helped |
| Add level-aware stretching (ablation) | HGCN · Disease node classification, dimension 64 | accuracy: 89.4±2.5 -> 95.0±0.8 | Helped |
| Add root alignment (ablation) | HGCN · Disease node classification, dimension 64 | accuracy: 94.5±0.6 -> 95.0±0.8 | Helped |
| Add level-aware stretching (ablation) | HGCN · Citeseer node classification, dimension 64 | accuracy: 66.7±0.9 -> 74.1±0.3 | Helped |
| Add root alignment (ablation) | HGCN · Citeseer node classification, dimension 64 | accuracy: 68.5±0.6 -> 74.1±0.3 | Helped |
| Push node embeddings toward the origin instead of away from it (ablation) | HGCN · Citeseer node classification, dimension 64 | accuracy: 74.1±0.3 -> 18.1±0.0 | Hurt |
| Use HC rather than graph centrality to identify the root (ablation) | HGCN with HIE · Disease node classification | accuracy: 85.0±2.9 -> 95.0±0.8 | Helped |
Closest papers by meaning
- Elastic Representations via Hyperbolic Geometry (NeurIPS 2026 · similarity 0.92)
- Unsupervised Hyperbolic Representation Learning via Message Passing Auto-Encoders (CVPR 2021 · similarity 0.91)
- Learning Along the Arrow of Time: Hyperbolic Geometry for Backward-Compatible Representation Learning (ICML 2025 · similarity 0.91)
- Hyperbolic Embeddings of Supervised Models (NeurIPS 2024 · similarity 0.90)
- Learning Structured Representations with Hyperbolic Embeddings (NeurIPS 2024 · similarity 0.88)
- Fully Hyperbolic Convolutional Neural Networks for Computer Vision (ICLR 2024 · 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.