Research map / Learning theory and algorithms / Prediction and decision losses
Ultrahyperbolic Representation Learning
Marc Law, Jos Stam
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 |
|---|---|---|---|
| use ultrahyperbolic Q^{3,1} instead of Euclidean R^4 | graph embedding · Zachary's karate club | rank of first leader: 5.4 -> 1.2 | Improved |
| use ultrahyperbolic Q^{3,1} instead of hyperbolic Q^{4,0} | graph embedding · Zachary's karate club | rank of first leader: 2.8 -> 1.2 | Improved |
| use ultrahyperbolic Q^{3,1} instead of spherical Q^{0,4} | graph embedding · Zachary's karate club | rank of second leader: 4.0 -> 2.4 | Improved |
| use ultrahyperbolic Q^{3,1} instead of hyperbolic Q^{4,0} | graph embedding · Zachary's karate club | top 5 Spearman's rho: 0.20 -> 0.76 | Improved |
| use ultrahyperbolic Q^{2,2} instead of hyperbolic Q^{4,0} | graph embedding · Zachary's karate club | top 10 Spearman's rho: 0.38 -> 0.79 | Improved |
| increase time dimensions to Q^{1,3} from Q^{3,1} (ablation) | graph embedding · Zachary's karate club | top 5 Spearman's rho: 0.76 -> 0.36 | Hurt |
| use Euclidean norm proxy instead of sum of distances for hyperbolic importance (ablation) | hyperbolic embedding · Zachary's karate club | rank of first leader: 2.8 -> 8.6 | Hurt |
| use Euclidean norm proxy instead of sum of distances for hyperbolic importance (ablation) | hyperbolic embedding · Zachary's karate club | rank of second leader: 4.2 -> 18.6 | Hurt |
| use pseudo-Riemannian optimizer instead of Euclidean-gradient mapping optimizer in low dimensions Q^{4,1}, Q^{4,2} (ablation) | ultrahyperbolic optimization · Zachary's karate club | satisfaction of all distance constraints | Improved |
| use -Df(x) instead of the projected direction -chi as search direction (ablation) | pseudo-Riemannian optimizer · Zachary's karate club | convergence | Hurt |
| use Euclidean-gradient mapping optimizer instead of pseudo-Riemannian optimizer (ablation) | ultrahyperbolic optimization · Zachary's karate club | training speed: 10% faster | Mixed |
Closest papers by meaning
- Ultrahyperbolic Neural Networks (NeurIPS 2021 · similarity 0.93)
- Fast Hyperboloid Decision Tree Algorithms (ICLR 2024 · similarity 0.88)
- Hyperbolic Embeddings of Supervised Models (NeurIPS 2024 · similarity 0.87)
- Weighted Embeddings for Low-Dimensional Graph Representation (AAAI 2025 · similarity 0.87)
- RMLR: Extending Multinomial Logistic Regression into General Geometries (NeurIPS 2024 · similarity 0.87)
- Robust Hyperbolic Learning with Curvature-Aware Optimization (NeurIPS 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.
