Research map / Federated learning and privacy / Differential privacy
Differentially Private Bayesian Linear Regression
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 |
|---|---|---|---|
| account for privacy-mechanism noise during posterior inference instead of treating noisy sufficient statistics as exact | Bayesian linear regression · synthetic data | posterior calibration | Helped |
| account for privacy-mechanism noise during posterior inference instead of treating noisy sufficient statistics as exact | Bayesian linear regression · synthetic data | utility (MMD to non-private posterior) | Helped |
| treat noisy sufficient statistics as exact | Naive · real-data predictive posterior | credible-interval coverage | Hurt |
| estimate covariate moments privately and do not infer the noise in those moment estimates (ablation) | Gibbs-SS-Noisy · synthetic data | posterior calibration | Hurt |
| derive covariate moments from a generic prior (ablation) | Gibbs-SS-Prior · synthetic data | posterior calibration | Mixed |
| instantiate latent variables for each individual | MCMC-Ind · increasing population size | runtime | Hurt |
| marginalize individuals and infer from sufficient statistics | Gibbs-SS variants · increasing population size | runtime scaling with population size | Helped |
| use privately estimated sample moments rather than requiring a specified covariate prior (ablation) | Gibbs-SS-Noisy · real-data predictive posterior | credible-interval coverage | Helped |
Closest papers by meaning
- Differentially Private Distributed Bayesian Linear Regression with MCMC (ICML 2023 · similarity 0.94)
- On Inherent Privacy of Posterior Sampling: A Unified R\'enyi-Divergence Framework (NeurIPS 2026 · similarity 0.91)
- Privacy Guarantees in Posterior Sampling under Contamination (NeurIPS 2026 · similarity 0.90)
- Private Gradient Descent for Linear Regression: Tighter Error Bounds and Instance-Specific Uncertainty Estimation (ICML 2024 · similarity 0.89)
- Truthful High Dimensional Sparse Linear Regression (NeurIPS 2024 · similarity 0.88)
- Privacy Amplification Through Synthetic Data: Insights from Linear Regression (ICML 2025 · 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.