Research map / Label-efficient and robust learning / Recognition and long-tail
Auto-Encoding Transformations in Reparameterized Lie Groups for Unsupervised Learning
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 |
|---|---|---|---|
| Use geodesic transformation loss with scaled-rotation projection instead of Euclidean matrix MSE | AET · CIFAR-10 | error rate: 9.41 -> 9.09 | Helped |
| Use geodesic transformation loss with scaled-rotation projection instead of Euclidean matrix MSE | AET · CIFAR-10 | error rate: 7.82 -> 7.44 | Helped |
| Use geodesic transformation loss with scaled-rotation projection instead of Euclidean matrix MSE | AET · ImageNet | Top-1 accuracy: 53.2 -> 54.3 (Conv4); 47.0 -> 47.5 (Conv5) | Helped |
| Use geodesic transformation loss with scaled-rotation projection instead of Euclidean matrix MSE | AET · Places | Top-1 accuracy: 22.1 -> 22.8 (Conv1); 32.9 -> 33.2 (Conv2); 37.1 -> 38.1 (Conv3); 36.2 -> 36.8 (Conv4); 34.7 -> 35.3 (Conv5) | Mixed |
| Use geodesic transformation loss with scaled-rotation projection instead of Euclidean matrix MSE | AET · ImageNet | Top-1 accuracy: 19.2 -> 19.6; 32.8 -> 34.1; 40.6 -> 41.9; 39.7 -> 40.4; 37.7 -> 37.9 | Helped |
| Use geodesic transformation loss with scaled-rotation projection instead of Euclidean matrix MSE | AET with ten-crop evaluation · ImageNet | Top-1 accuracy: 19.3 -> 21.2; 35.4 -> 36.9; 44.0 -> 45.9; 43.6 -> 44.7; 42.4 -> 43.2 | Helped |
Closest papers by meaning
- Lie Group Decompositions for Equivariant Neural Networks (ICLR 2024 · similarity 0.87)
- Neural Isometries: Taming Transformations for Equivariant ML (NeurIPS 2024 · similarity 0.87)
- Learning Infinitesimal Generators of Continuous Symmetries from Data (NeurIPS 2024 · similarity 0.86)
- Building Transformation Layers for Riemannian Neural Networks (NeurIPS 2026 · similarity 0.85)
- Connecting Neural Models Latent Geometries with Relative Geodesic Representations (NeurIPS 2025 · similarity 0.85)
- Graph Geometry-Preserving Autoencoders (ICML 2024 · similarity 0.85)
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.