Research map / Learning theory and algorithms / Sample complexity
Sharp uniform convergence bounds through empirical centralization
Cyrus Cousins, Matteo Riondato
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 empirical centralization of the function family instead of non-centralized bounds | Supremum deviation bounds · Synthetic distributions in R^d, m up to 10^7, delta=0.01 | SD bound tightness | Improved |
| Use empirical centralization at very small sample sizes (all bounds vacuous) | Supremum deviation bounds · Synthetic distributions, small m | SD bound tightness | Hurt |
| Monte-Carlo estimate of ERA with n=32 trials instead of analytical ERA bound | Centralized SD bound · Synthetic distributions | SD bound tightness | Improved |
| Increase dimension d from 4 to 256 for family F1 | MC-ERA bound · F1 experiments | MC-ERA magnitude | Hurt |
| Increase number of Monte-Carlo trials n | MC-ERA to ERA gap · Synthetic distributions | bound gap | Improved |
Closest papers by meaning
- Relative Deviation Margin Bounds (ICML 2021 · similarity 0.87)
- Time-uniform confidence bands for the CDF under nonstationarity (NeurIPS 2023 · similarity 0.86)
- A General-Purpose Theorem for High-Probability Bounds of Stochastic Approximation with Polyak Averaging (NeurIPS 2025 · similarity 0.85)
- The Empirical Mean is Minimax Optimal for Local Glivenko-Cantelli (ICML 2025 · similarity 0.85)
- Exactly Tight Information-theoretic Generalization Bounds via Binary Jensen-Shannon Divergence (ICML 2025 · similarity 0.84)
- Robust Statistical Estimators with Bounded Empirical Sensitivity (NeurIPS 2026 · 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.
