Research map / Federated learning and privacy / Differential privacy
Differentially Private Hypothesis Testing for Linear Regression
Daniel G. Alabi, Salil P. Vadhan
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 DP bootstrap confidence interval test instead of DP F-statistic | DP linear relationship tester · Synthetic data (normal/uniform X) | power | Hurt |
| use DP Bernoulli tester instead of DP F-statistic | DP linear relationship tester · Synthetic data | power | Mixed |
| increase true slope from 0.1 to 1 | DP F-statistic linear test · Synthetic data | power: slopes 0.1, 1 | Improved |
| decrease noise sigma_e of dependent variable to 0.001 | DP F-statistic linear test · Synthetic data | power: sigma_e values 0.001, 0.35, 1 | Hurt |
| use unbalanced group split (1/8 vs 7/8) in mixture test | DP F-statistic mixture test · Synthetic data | significance level: fractions 1/8, 1/4, 1/2 | Hurt |
| change distribution of X (normal, uniform, exponential) with equal variance | DP F-statistic linear test · Synthetic data | significance level | No clear effect |
Closest papers by meaning
- Hypothesis Testing for Differentially Private Linear Regression (NeurIPS 2022 · similarity 0.98)
- Differentially private methods for managing model uncertainty in linear regression (JMLR 2024 · similarity 0.89)
- Differentially Private Estimation and Inference in High-Dimensional Regression with FDR Control (JMLR 2026 · similarity 0.88)
- Differentially Private Bayesian Linear Regression (NeurIPS 2019 · similarity 0.88)
- The Test of Tests: A Framework for Differentially Private Hypothesis Testing (ICML 2023 · similarity 0.87)
- Priv’IT: Private and Sample Efficient Identity Testing (ICML 2017 · 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.
