Research map / Learning theory and algorithms / Kernels and regression theory
Implicit Regularization of Random Feature Models
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 a finite number of Gaussian random features with ridge λ instead of kernel ridge regression with ridge λ | Gaussian random feature models · Fixed training dataset | effective ridge | Helped |
| Increase the number of Gaussian random features | Gaussian random feature models · Fixed training dataset | effective ridge | Helped |
| Approximate the finite-feature predictor with KRR using its effective ridge λ̃ | Expected Gaussian random feature predictor · MNIST experiments | test MSE | No clear effect |
| Use random Fourier features instead of Gaussian features | Random feature predictors · MNIST experiments | test error | Hurt |
| Increase the feature-to-data ratio | Expected random Fourier feature predictor · MNIST experiments | difference in test errors from effective-ridge KRR | Helped |
| Use effective-ridge KRR to approximate the random Fourier feature predictor when P ≥ N | Random Fourier feature predictor · MNIST experiments | test error | No clear effect |
| Use effective-ridge KRR to approximate the random Fourier feature predictor | Random Fourier feature predictor · MNIST, N=1000 | test error | No clear effect |
| Use ridgeless Gaussian random features with P ≥ N | Gaussian random feature predictor · Overparameterized regime | predictor bias relative to ridgeless KRR | No clear effect |
| Operate near the interpolation threshold γ=1 with small ridge | Gaussian random feature predictor · Interpolation threshold | test error | Hurt |
Closest papers by meaning
- Optimal Kernel Quantile Learning with Random Features (ICML 2024 · similarity 0.89)
- Dimension-free deterministic equivalents and scaling laws for random feature regression (NeurIPS 2024 · similarity 0.88)
- Towards Theoretical Understanding of Learning Large-scale Dependent Data via Random Features (ICML 2024 · similarity 0.87)
- Kernel Regression in Structured Non-IID Settings: Theory and Implications for Denoising Score Learning (NeurIPS 2025 · similarity 0.87)
- A Comprehensive Analysis on the Learning Curve in Kernel Ridge Regression (NeurIPS 2024 · similarity 0.87)
- Optimal Ridge Regularization for Out-of-Distribution Prediction (ICML 2024 · 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.