Research map / Learning theory and algorithms / Prediction and decision losses
Iterative Approximate Cross-Validation
Yuetian Luo, Zhimei Ren, Rina Barber
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 |
|---|---|---|---|
| Use IACV instead of approximate Newton or infinitesimal jackknife during early iterations | GD logistic regression with ridge · Logistic regression with ridge, GD | approximation error | Improved |
| Use IACV instead of approximate Newton or infinitesimal jackknife during early iterations | GD logistic regression with ridge · Logistic regression with ridge, GD | relative CV error | Improved |
| Use IACV instead of exact leave-one-out CV | GD logistic regression with ridge · Logistic regression with ridge, n=1000 | runtime speedup: approximately 6-7 times speed-up | Improved |
| Increase sample size from 250 to 1000 | IACV with GD logistic regression and ridge · Logistic regression with ridge, GD | limiting approximation error: 1.5 \times 10^{-3} to 6.8 \times 10^{-5} | Improved |
| Use IACV instead of approximate Newton or infinitesimal jackknife during early iterations | SGD logistic regression with ridge · Logistic regression with ridge, SGD | approximation error | Improved |
| Use IACV instead of approximate Newton or infinitesimal jackknife during early iterations | SGD logistic regression with ridge · Logistic regression with ridge, SGD | CV error | Improved |
| Use IACV instead of exact leave-one-out CV | SGD logistic regression with ridge · Logistic regression with ridge, SGD | runtime | Improved |
| Use IACV instead of approximate Newton or infinitesimal jackknife during early iterations | ProxGD logistic regression with L1 regularization · Logistic regression with L1 regularizer, ProxGD | approximation error | Improved |
| Use IACV instead of approximate Newton or infinitesimal jackknife during early iterations | ProxGD logistic regression with L1 regularization · Logistic regression with L1 regularizer, ProxGD | CV error | Improved |
| Replace IACV with approximate Newton before convergence | GD and ProxGD · Early optimization iterations | approximation error | Hurt |
Closest papers by meaning
- Fast Cross-Validation for Kernel-based Algorithms (TPAMI 2019 · similarity 0.90)
- Approximate Cross-Validation for Structured Models (NeurIPS 2020 · similarity 0.88)
- Fast and Informative Model Selection Using Learning Curve Cross-Validation (TPAMI 2023 · similarity 0.87)
- Approximate Cross-Validation with Low-Rank Data in High Dimensions (NeurIPS 2020 · similarity 0.87)
- Is Cross-validation the Gold Standard to Estimate Out-of-sample Model Performance? (NeurIPS 2024 · similarity 0.84)
- Crossing the Validation crisis: Cross-validation reduces benchmarking variance surprisingly well (NeurIPS 2026 · similarity 0.83)
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.
