Research map / Learning theory and algorithms / Kernels and regression theory
Generalisation error in learning with random features and the hidden manifold model
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 logistic loss instead of square loss | Random Gaussian features with optimal regularisation · Kernel limit | generalisation error: 0.17 -> 0.16 | Helped |
| Use orthogonal projections instead of Gaussian random projections | Random features · Regression and classification | generalisation error | Helped |
| Use orthogonal projections instead of Gaussian random projections | Random features · Logistic classification | linear separability threshold | Helped |
| Increase the number of features beyond the logistic interpolation threshold | Logistic regression with random Gaussian features · Classification | generalisation error | Mixed |
| Use logistic loss instead of square loss | Random Gaussian features · Classification, except near the logistic interpolation peak | generalisation error | Helped |
| Increase ridge regularisation | Hidden manifold model with square loss · Classification near the interpolation threshold | generalisation error | Helped |
| Decrease the latent-to-data dimension ratio | Hidden manifold model with optimally regularised square loss · Classification | generalisation error | Helped |
| Use more features per sample | Orthogonal random features · Regression and classification | features needed to reach the kernel limit | Helped |
Closest papers by meaning
- Fluctuations, Bias, Variance & Ensemble of Learners: Exact Asymptotics for Convex Losses in High-Dimension (ICML 2022 · similarity 0.88)
- Diffusion Models and the Manifold Hypothesis: Log-Domain Smoothing is Geometry Adaptive (NeurIPS 2025 · similarity 0.86)
- Generalization error of spectral algorithms (ICLR 2024 · similarity 0.86)
- Manifold Random Features (NeurIPS 2026 · similarity 0.86)
- Double-Descent Curves in Neural Networks: A New Perspective Using Gaussian Processes (AAAI 2024 · similarity 0.86)
- Generalization in Kernel Regression Under Realistic Assumptions (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.