Research map / Learning theory and algorithms / Kernels and regression theory
Diffeomorphic Learning
Laurent Younes
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 diffeomorphic transformation of the data | logistic regression · Tori (d=10, 250 per class) | test error rate: 0.325 -> 0.023 | Improved |
| add diffeomorphic transformation of the data | logistic regression · Spherical layers (d=3, 100 per class) | test error rate: 0.507 -> 0.246 | Improved |
| add diffeomorphic transformation of the data | logistic regression · Xor (d=50, 100 per class) | test error rate: 0.505 -> 0.066 | Improved |
| add diffeomorphic transformation of the data | logistic regression · MNIST subset (d=196, 100 per class) | test error rate: 0.128 -> 0.113 | Improved |
| add diffeomorphic transformation of the data | kernel SVM · MNIST subset (d=196, 100 per class) | test error rate: 0.228 -> 0.325 | Hurt |
| add diffeomorphic transformation of the data | MLP5 · RBF data (d=2, 500 per class) | test error rate: 0.081 -> 0.142 | Hurt |
| add diffeomorphic transformation of the data | MLP1 · Segment lengths (d=100, 100 per class) | test error rate: 0.337 -> 0.390 | Hurt |
| add diffeomorphic transformation of the data | logistic regression · Segment pairs (d=50, 100 per class) | test error rate: 0.477 -> 0.461 | No clear effect |
| add diffeomorphic transformation of the data | logistic regression · Tori (d=20, 100 per class) | test error rate: 0.320 -> 0.321 | No clear effect |
| increase time discretization steps from T=1 to T=100 (ablation) | Diffeomorphic learning (logistic regression on transformed data) · Xor (100 per class) | test error rate: 0.478 -> 0.039 | Improved |
| increase time discretization steps from T=1 to T=100 (ablation) | Diffeomorphic learning (logistic regression on transformed data) · Mixture of Gaussians (100 per class) | test error rate: 0.178 -> 0.109 | Improved |
| use kernel scale 4 rho_0 instead of rho_0 (ablation) | Diffeomorphic learning (logistic regression on transformed data) · Xor (100 per class) | test error rate: 0.092 -> 0.01 | Improved |
Closest papers by meaning
- FineMorphs: Affine-Diffeomorphic Sequences for Regression (JMLR 2024 · similarity 0.89)
- Diffeomorphism-based feature learning using Poincaré inequalities on augmented input space (JMLR 2025 · similarity 0.88)
- Deep Diffeomorphic Transformer Networks (CVPR 2018 · similarity 0.87)
- Parametric Surface Diffeomorphometry for Low Dimensional Embeddings of Dense Segmentations and Imagery (TPAMI 2016 · similarity 0.86)
- Denoising Functional Maps: Diffusion Models for Shape Correspondence (CVPR 2025 · similarity 0.86)
- Topology-Preserving Shape Reconstruction and Registration via Neural Diffeomorphic Flow (CVPR 2022 · 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.
