Research map / Learning theory and algorithms / Sample complexity
Faster Algorithms and Constant Lower Bounds for the Worst-Case Expected Error
Jonah Brown-Cohen
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 the ℓ2 algorithm instead of subgroup estimation | Importance sampling · Constant and Intergroup Variance | expected squared error: 0.018 -> 0.052 | Hurt |
| Use the ℓ2 algorithm instead of reweighting | Importance sampling · Intragroup Variance | expected squared error: 0.100 -> 0.053 | Improved |
| Use the ℓ2 algorithm instead of reweighting | Importance sampling · Worst Case ℓ∞ and Worst Case ℓ2 | expected squared error: 0.101 -> 0.062; 0.181 -> 0.078 | Improved |
| Use the ℓ2 algorithm instead of the sample mean | Snowball sampling · Spatially Correlate Values | expected squared error: 0.082 -> 0.032 | Improved |
| Use the ℓ2 algorithm instead of the ℓ∞ algorithm | Snowball sampling · Worst Case ℓ∞ | expected squared error: 0.135 -> 0.153 | Hurt |
| Use the ℓ2 algorithm instead of the ℓ∞ algorithm | Snowball sampling · Worst Case ℓ2 | expected squared error: 0.327 -> 0.326 | Improved |
| Use the ℓ2 algorithm instead of the ℓ∞ algorithm | Selective prediction · Worst Case ℓ∞ | expected squared error: 0.498 -> 0.620 | Hurt |
| Use the ℓ2 algorithm instead of the ℓ∞ algorithm | Selective prediction · Worst Case ℓ2 | expected squared error: 0.746 -> 0.686 | Improved |
| Use the ℓ2 algorithm rather than the original ℓ∞ algorithm | Importance sampling · Synthetic datasets | expected squared error: 0.051 -> 0.052; 0.053 -> 0.052; 0.052 -> 0.053 | No clear effect |
Closest papers by meaning
- Optimality in Mean Estimation: Beyond Worst-Case, Beyond Sub-Gaussian, and Beyond $1+\alpha$ Moments (NeurIPS 2023 · similarity 0.88)
- Worst-Case Analysis for Randomly Collected Data (NeurIPS 2020 · similarity 0.87)
- Simple and Optimal Sublinear Algorithms for Mean Estimation (NeurIPS 2025 · similarity 0.85)
- One Sample Fits All: Approximating All Probabilistic Values Simultaneously and Efficiently (NeurIPS 2024 · similarity 0.84)
- Flat Minima in Linear Estimation and an Extended Gauss Markov Theorem (ICLR 2024 · similarity 0.84)
- Sub-Gaussian Confidence Intervals for Heavy-Tailed Data: Characterizing the Limits of Inference (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.
