Research map / Learning theory and algorithms / Sample complexity
Subset-Based Instance Optimality in Private Estimation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Subset-based benchmark (removal-only, large subsets) instead of add/remove local minimax risk for mean estimation | private mean estimation · theoretical comparison | instance-dependent risk | Helped |
| Tighter clipping threshold choice (rank targets t_l, t_u balancing bias and variance) in the DP mean algorithm | private mean estimation · theoretical analysis | expected error vs. lower bound | Helped |
| Subset-optimal mean algorithm matches or exceeds prior algorithms' asymptotic rates under sub-Gaussian and bounded-moment assumptions | statistical mean estimation · sub-Gaussian, bounded k-th moment families | worst-case expected error rate | No clear effect |
| Pruning outliers vs clipping in monotone-property algorithm, with added discretization | general monotone properties · theory | expected loss | No clear effect |
Closest papers by meaning
- Instance-Optimality for Private KL Distribution Estimation (NeurIPS 2025 · similarity 0.91)
- Instance-Specific Asymmetric Sensitivity in Differential Privacy (NeurIPS 2024 · similarity 0.90)
- Locally Optimal Private Sampling: Beyond the Global Minimax (NeurIPS 2025 · similarity 0.89)
- Lightweight Protocols for Distributed Private Quantile Estimation (ICML 2025 · similarity 0.89)
- Public-data Assisted Private Stochastic Optimization: Power and Limitations (NeurIPS 2024 · similarity 0.89)
- Exactly Minimax-Optimal Locally Differentially Private Sampling (NeurIPS 2024 · similarity 0.88)
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.