Research map / Learning theory and algorithms / Kernels and regression theory
Boosted Kernel Ridge Regression: Optimal Learning Rates and Early Stopping
Shao-Bo Lin, Yunwen Lei, Ding-Xuan Zhou
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 |
|---|---|---|---|
| increase regularization parameter lambda from 0.0004 to 0.0128 (ablation) | BKRR · Regression g1 | excess generalization error: 2.72e-4 -> 1.87e-4 | Improved |
| increase regularization parameter lambda from 0.0004 to 0.0128 (ablation) | BKRR · Regression g2 | excess generalization error: 1.76e-3 -> 1.60e-3 | Improved |
| use larger lambda beyond about 1e-2 with tuned iterations (ablation) | BKRR · Regression g1 and g2 | excess generalization error: 1.87e-4 -> 1.87e-4 | No clear effect |
| select number of iterations in BKRR instead of lambda in ITR | BKRR vs ITR · Regression g1 and g2 | parameter selection stability | Improved |
| use BKRR instead of ITR | BKRR · Regression g1 | excess generalization error: 1.87e-4 -> 1.87e-4 | No clear effect |
| use BKRR instead of ITR | BKRR · Regression g2 | excess generalization error: 1.59e-3 -> 1.59e-3 | No clear effect |
| increase lambda from 0.0032 to 0.2048 (ablation) | BKRR · Regression g1 and g2 | iteration number at minimum EGE | Improved |
| continue boosting iterations past the optimum (ablation) | BKRR · Regression g1 and g2 | excess generalization error | Hurt |
| use adaptive stopping rule instead of five-fold CV | BKRR · Regression g1, sample sizes 800-4000 | excess generalization error | Improved |
| use adaptive stopping rule instead of five-fold CV | BKRR · Regression g1 | training cost | Improved |
| use adaptive stopping rule | BKRR · Regression g1 | excess generalization error vs oracle | Improved |
Closest papers by meaning
- A Comprehensive Analysis on the Learning Curve in Kernel Ridge Regression (NeurIPS 2024 · similarity 0.90)
- Kernel Regression in Structured Non-IID Settings: Theory and Implications for Denoising Score Learning (NeurIPS 2025 · similarity 0.89)
- Provable and Efficient Dataset Distillation for Kernel Ridge Regression (NeurIPS 2024 · similarity 0.87)
- Supervised Kernel Thinning (NeurIPS 2024 · similarity 0.87)
- Statistical Inference for Gradient Boosting Regression (NeurIPS 2025 · similarity 0.86)
- Characterizing Overfitting in Kernel Ridgeless Regression Through the Eigenspectrum (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.
