Research map / Federated learning and privacy / Differential privacy
Smoothly Bounding User Contributions in Differential Privacy
Alessandro Epasto, Mohammad Mahdian, Jieming Mao, Vahab Mirrokni, Lijie Ren
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 weighted averaging instead of sample limiting with best threshold h* | private linear regression · drugs, eps=1 | average squared error: 24.8 -> 3.1 | Improved |
| use weighted averaging instead of sample limiting with best threshold h* | private linear regression · drugs, eps=2 | average squared error: 7.7 -> 2.5 | Improved |
| use weighted averaging instead of sample limiting with best threshold h* | private linear regression · drugs, eps=3 | average squared error: 4.5 -> 2.3 | Improved |
| use weighted averaging instead of sample limiting with best threshold h* | private linear regression · news, eps=1 | average squared error: 96344.4 -> 1696.3 | Improved |
| use weighted averaging instead of sample limiting with best threshold h* | private linear regression · news, eps=2 | average squared error: 24110.0 -> 440.1 | Improved |
| use weighted averaging instead of sample limiting with best threshold h* | private linear regression · news, eps=3 | average squared error: 10648.3 -> 166.2 | Improved |
| increase epsilon from 1 to 3 | weighted averaging · news | average squared error: 1696.3 -> 166.2 | Improved |
| increase epsilon from 1 to 3 | weighted averaging · drugs | average squared error: 3.1 -> 2.3 | Improved |
| use weighted averaging instead of using all data (h=max) | private linear regression · news, eps=1 | average squared error: 4862670.7 -> 1696.3 | Improved |
Closest papers by meaning
- Privacy amplification by random allocation (NeurIPS 2025 · similarity 0.84)
- Certification for Differentially Private Prediction in Gradient-Based Training (ICML 2025 · similarity 0.84)
- Provable Privacy with Non-Private Pre-Processing (ICML 2024 · similarity 0.83)
- Privacy Guarantees in Posterior Sampling under Contamination (NeurIPS 2026 · similarity 0.83)
- Faster Algorithms for User-Level Private Stochastic Convex Optimization (NeurIPS 2024 · similarity 0.83)
- Setting $\varepsilon$ is not the Issue in Differential Privacy (NeurIPS 2025 · similarity 0.83)
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.
