Research map / Probabilistic and scientific modeling / Variational and generative sampling
A Rotated Hyperbolic Wrapped Normal Distribution for Hierarchical Representation Learning
Seunghyuk Cho, Juyong Lee, Jaesik Park, Dongwoo Kim
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 |
|---|---|---|---|
| Rotate the covariance toward the mean direction rather than use full-covariance HWN | Hyperbolic VAE · Noisy synthetic binary tree, depth 5 | Correlation w/ depth: 0.878_{\pm 0.051} -> 0.911_{\pm 0.027} | Improved |
| Rotate the covariance toward the mean direction rather than use diagonal HWN | Hyperbolic VAE · Noisy synthetic binary tree, depth 6 | Correlation w/ distance: 0.817_{\pm 0.010} -> 0.822_{\pm 0.017} | No clear effect |
| Use RoWN rather than full-covariance HWN | Probabilistic hyperbolic word embedding · WordNet, latent dimension 10 | MR: 466.513_{\pm 75.142} -> 2.888_{\pm 0.162} | Improved |
| Use RoWN rather than diagonal HWN | Probabilistic hyperbolic word embedding · WordNet, latent dimension 20 | mAP: 0.905_{\pm 0.007} -> 0.921_{\pm 0.005} | Improved |
| Use RoWN rather than isotropic HWN | Hyperbolic VAE · Atari 2600 Breakout, latent dimension 20 | Correlation btw. score and norm: 0.607_{\pm 0.015} -> 0.561_{\pm 0.029} | Hurt |
| Learn the rotation direction instead of fixing it toward the mean (ablation) | RoWN · Noisy synthetic binary tree, depth 7 | Correlation w/ depth: 0.827_{\pm 0.047} -> 0.711_{\pm 0.072} | Hurt |
| Learn the rotation direction instead of fixing it toward the mean (ablation) | RoWN · Noisy synthetic binary tree, depth 4 | Correlation w/ depth: 0.930_{\pm 0.026} -> 0.952_{\pm 0.018} | Improved |
Closest papers by meaning
- A Wrapped Normal Distribution on Hyperbolic Space for Gradient-Based Learning (ICML 2019 · similarity 0.91)
- Hyperbolic Embeddings of Supervised Models (NeurIPS 2024 · similarity 0.87)
- Learning Structured Representations with Hyperbolic Embeddings (NeurIPS 2024 · similarity 0.85)
- Hyperbolic Graph Diffusion Model (AAAI 2024 · similarity 0.85)
- Hyperbolic Dataset Distillation (NeurIPS 2025 · similarity 0.84)
- Fully Hyperbolic Convolutional Neural Networks for Computer Vision (ICLR 2024 · similarity 0.84)
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.
