Research map / Federated learning and privacy / Differential privacy
Priv’IT: Private and Sample Efficient Identity Testing
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 |
|---|---|---|---|
| Combine noisy-count filtering with a non-noised chi-square statistic whose sensitivity is bounded after filtering | Priv'IT · Private identity testing | sample complexity: m = \tilde O\left(\max \left\{\frac{\sqrt{n}}{\alpha^2}, \frac{\sqrt{n}}{\alpha^{3/2}\varepsilon}, \frac{n^{1/3}}{\alpha | Helped |
| Add Laplace noise to counts before applying the chi-square tester | Identity testing · Uniformity testing | sample complexity: Ω(n^{3/4}) | Hurt |
| Use repetition to privatize a non-private tester | Identity testing · Identity testing | sample complexity: O\left(\frac{\sqrt{n}}{\varepsilon\alpha^2} \cdot \log(1/\beta)\right) | Hurt |
| Use Priv'IT instead of zCDP-GOF | Identity testing · Uniformity testing with Paninski alternatives, pure-DP-equivalent zCDP guarantee | minimum samples | Hurt |
| Use Priv'IT instead of zCDP-GOF | Identity testing · 2-histogram with low-probability support elements | minimum samples | Helped |
| Use MCGOF on a 2-histogram with rare elements | MCGOF · Identity testing on a 2-histogram, n=400 | samples: 130000 | Hurt |
| Use Priv'IT instead of zCDP-GOF | Identity testing · Uniformity testing under approximate-DP comparisons | minimum samples | Helped |
| Use Priv'IT instead of MCGOF | Identity testing · Runtime comparison | support size reached in the same amount of time: 20000 vs 10000 | Helped |
| Use zCDP-GOF instead of Priv'IT | Identity testing · Memory comparison | support size before running out of memory: 11800 | Hurt |
Closest papers by meaning
- Minimax Optimal Two-Sample Testing under Local Differential Privacy (NeurIPS 2026 · similarity 0.91)
- Sort, Partition, Randomize: Optimal Binary Hypothesis Testing under Local Differential Privacy (NeurIPS 2026 · similarity 0.89)
- Optimal Rates for Differentially Private Hypothesis Testing with E-values (NeurIPS 2026 · similarity 0.89)
- Privacy Guarantees in Posterior Sampling under Contamination (NeurIPS 2026 · similarity 0.87)
- Differentially Private Equivalence Testing for Continuous Distributions and Applications (NeurIPS 2024 · similarity 0.87)
- Locally Optimal Private Sampling: Beyond the Global Minimax (NeurIPS 2025 · similarity 0.87)
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.