Research map / Federated learning and privacy / Differential privacy
An Uncertainty Principle is a Price of Privacy-Preserving Microdata
John M. Abowd, Robert Ashmead, Ryan Cumings-Menon, Simson L. Garfinkel, Daniel Kifer, Philip Leclerc, William Sexton, Ashley E Simpson, Christine Task, Pavel Zhuravlev
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 ReWeight instead of NNLS postprocessing | Laplace mechanism microdata · Sum query, D01 | squared error: 461.9 -> 108.5 | Improved |
| use Sequential fitting instead of NNLS | Laplace mechanism microdata · Sum query, D01 | squared error: 461.9 -> 149.2 | Improved |
| use MaxFit instead of OLS (nonnegative microdata) | Laplace mechanism · Sum query, D06 | squared error: 107.2 -> 677.4 | Hurt |
| use NNLS instead of OLS (nonnegative microdata) | Laplace mechanism · Sum query, D06 | squared error: 107.2 -> 584.0 | Hurt |
| use ReWeight instead of NNLS | Laplace mechanism microdata · Sum query, D13 | squared error: 300.3 -> 108.8 | Improved |
| use ReWeight instead of NNLS | Laplace mechanism microdata · Identity queries total, D01 | total squared error: 344.2 -> 159.2 | Improved |
| use ReWeight instead of NNLS | Laplace mechanism microdata · Identity queries max cell error, D08 | max squared error: 130.0 -> 168.9 | Hurt |
| use ReWeight instead of NNLS | Laplace mechanism microdata · Identity queries total, D13 | total squared error: 2884.1 -> 2870.4 | No clear effect |
| use Sequential instead of ReWeight | Laplace mechanism microdata · Identity queries total, D13 | total squared error: 2870.4 -> 2484.2 | Improved |
| use ReWeight instead of Sequential | Laplace mechanism microdata · Sum query, D01 | squared error: 149.2 -> 108.5 | Improved |
| use ReWeight instead of NNLS | Laplace mechanism microdata · Identity queries max cell error, D01 | max squared error: 147.4 -> 78.4 | Improved |
| use ReWeight instead of NNLS | Laplace mechanism microdata · Identity queries max cell error, D13 | max squared error: 119.2 -> 166.1 | Hurt |
Closest papers by meaning
- A Central Limit Theorem for Differentially Private Query Answering (NeurIPS 2021 · similarity 0.90)
- Accuracy First: Selecting a Differential Privacy Level for Accuracy Constrained ERM (NeurIPS 2017 · similarity 0.88)
- Differential Privacy for Bayesian Inference through Posterior Sampling (JMLR 2017 · similarity 0.88)
- Private Query Release Assisted by Public Data (ICML 2020 · similarity 0.88)
- Privacy Guarantees in Posterior Sampling under Contamination (NeurIPS 2026 · similarity 0.87)
- Optimality of Matrix Mechanism on $\ell_p^p$-metric (ICLR 2025 · similarity 0.86)
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.
