Research map / Federated learning and privacy / Differential privacy
Differentially Private Aggregation in the Shuffle Model: Almost Central Accuracy in Almost a Single Message
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 |
|---|---|---|---|
| Use correlated zero-sum noise atoms with an integer change-of-basis analysis | Δ-summation in the shuffle model · Δ-summation | MSE: at most that of the Discrete Laplace mechanism with parameter (1 - γ)ε/Δ | Helped |
| Use correlated zero-sum noise atoms with an integer change-of-basis analysis | Δ-summation in the shuffle model · Δ-summation | messages per user: 1 + O(Δ log(1/δ)/(γεn)) + O(Δ log^2 Δ · log(Δ/δ)/(γεn)) | Helped |
| Set ε* to 0.9ε | Correlated Noise · Δ-summation | RMSE: 0.9 ε | Hurt |
| Use IKOS split-and-mix instead of Correlated Noise | Δ-summation · Δ-summation | RMSE | No clear effect |
| Use fragmented RAPPOR instead of Correlated Noise | Δ-summation · Δ-summation | RMSE: Θ(Δ^{3/2}) | Hurt |
| Increase n while holding Δ fixed | Correlated Noise · Δ-summation | messages per user: 1 + o(1) | Helped |
| Increase n | RAPPOR · Δ-summation | RMSE: less than 1% | No clear effect |
| Use randomized discretization and Δ-summation for real-valued inputs | Correlated Noise · real summation | MSE: at most the MSE of the Laplace mechanism with parameter (1 - ζ)ε | Helped |
Closest papers by meaning
- Private Vector Mean Estimation in the Shuffle Model: Optimal Rates Require Many Messages (ICML 2024 · similarity 0.91)
- A Generalized Shuffle Framework for Privacy Amplification: Strengthening Privacy Guarantees and Enhancing Utility (AAAI 2024 · similarity 0.91)
- Privacy amplification by random allocation (NeurIPS 2025 · similarity 0.88)
- Learning from End User Data with Shuffled Differential Privacy over Kernel Densities (ICLR 2025 · similarity 0.88)
- Sort, Partition, Randomize: Optimal Binary Hypothesis Testing under Local Differential Privacy (NeurIPS 2026 · similarity 0.87)
- Differentially Private Decentralized Learning with Random Walks (ICML 2024 · 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.