Research map / Learning theory and algorithms / Sample complexity
Choice of V for V-Fold Cross-Validation in Least-Squares Density Estimation
Sylvain Arlot, Matthieu Lerasle
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 V from 2 to 5 | V-fold penalization · Setting L, Dya2 | ratio to oracle risk: 10.21 -> 7.47 | Improved |
| increase V from 2 to 5 | V-fold penalization · Setting S, Dya2 | ratio to oracle risk: 2.39 -> 2.16 | Improved |
| increase V from 5 to 10 | V-fold penalization · Setting L, Dya2 | ratio to oracle risk: 7.47 -> 6.89 | Improved |
| increase V from 10 to n=500 | V-fold penalization · Setting L, Dya2 | ratio to oracle risk: 6.89 -> 6.35 | Improved |
| increase V from 2 to 10 | V-fold cross-validation · Setting L, Dya2 | ratio to oracle risk: 6.41 -> 6.24 | No clear effect |
| use V-fold penalty instead of V-fold CV at V=2 | model selection · Setting L, Dya2 | ratio to oracle risk: 6.41 -> 10.21 | Hurt |
| use V-fold penalty instead of V-fold CV at V=2 | model selection · Setting S, Dya2 | ratio to oracle risk: 2.05 -> 2.39 | Hurt |
| use V-fold penalty with V=n instead of pen_dim | model selection · Setting L, Dya2 | ratio to oracle risk: 8.27 -> 6.35 | Improved |
| use V-fold penalty with V=n instead of pen_dim | model selection · Setting S, Dya2 | ratio to oracle risk: 3.21 -> 2.06 | Improved |
| overpenalize V-fold penalty with factor C near 2.5 (ablation) | penLOO · Setting L, Dya2 | ratio to oracle risk: C=2.5 for L-Dya2; C=1.4 for S-Dya2 | Improved |
| overpenalize V-fold penalty with factor C near 1.4 (ablation) | penLOO · Setting S, Dya2 | ratio to oracle risk: C=1.4 for S-Dya2 | Improved |
| use Dya2 collection instead of regular histograms (Regu) (ablation) | oracle estimator · Setting L | quadratic risk x10^3: 13.4 -> 5.46 | Improved |
Closest papers by meaning
- Is Cross-validation the Gold Standard to Estimate Out-of-sample Model Performance? (NeurIPS 2024 · similarity 0.86)
- Crossing the Validation crisis: Cross-validation reduces benchmarking variance surprisingly well (NeurIPS 2026 · similarity 0.85)
- Asymptotically Free Sketched Ridge Ensembles: Risks, Cross-Validation, and Tuning (ICLR 2024 · similarity 0.84)
- Risk and cross validation in ridge regression with correlated samples (ICML 2025 · similarity 0.84)
- Sobolev Space Regularised Pre Density Models (ICML 2024 · similarity 0.84)
- Improving the Statistical Efficiency of Cross-Conformal Prediction (ICML 2025 · similarity 0.84)
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.
