Research map / Federated learning and privacy / Differential privacy
An Algorithmic Framework For Differentially Private Data Analysis on Trusted Processors
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 oblivious sorting and add Laplace noise | Distinct-elements estimation · Distinct elements | additive error: 1/\epsilon | Helped |
| Maintain a streaming sketch and add Laplace noise | Distinct-elements estimation · Distinct elements | space: \frac{\log n}{\alpha^2} + (\log n)^2 | Helped |
| Use streaming sketching instead of exact oblivious sorting | Distinct-elements estimation · Distinct elements | approximation error: (1 \pm \alpha)n^* \pm \log(1/\theta) \frac{1}{\epsilon} | Mixed |
| Add fake records, dummy records, and an oblivious shuffle to the noisy histogram computation | Histogram computation · Histogram | maximum per-bin absolute error: \log (k/\theta) \cdot \frac{2}{\epsilon} | Helped |
| Use the ODP histogram construction instead of an ORAM-based transformation | Histogram computation · Histogram | external memory accesses: O(n \log k) -> O(\tilde{n} \frac{\log \tilde{n}}{\log \log \tilde{n}}) | Helped |
| Obliviously sort item counts, add Laplace noise, then report the top k | Heavy-hitter detection · Heavy hitters | per-item absolute error: \sqrt{ n \cdot \log (n/\theta) \cdot \frac{\log m}{\epsilon}} -> \log (m/\theta) \cdot \frac{2}{\epsilon} | Helped |
| Keep the entire count-min sketch in private memory and add Laplace noise before release | Frequency oracle · Frequency oracle | memory access privacy | Helped |
Closest papers by meaning
- Distributed Differentially Private Data Analytics via Secure Sketching (ICML 2025 · similarity 0.89)
- Provable Privacy with Non-Private Pre-Processing (ICML 2024 · similarity 0.89)
- A Generalized Shuffle Framework for Privacy Amplification: Strengthening Privacy Guarantees and Enhancing Utility (AAAI 2024 · similarity 0.87)
- Auditing $f$-differential privacy in one run (ICML 2025 · similarity 0.87)
- Setting $\varepsilon$ is not the Issue in Differential Privacy (NeurIPS 2025 · similarity 0.87)
- Learning from End User Data with Shuffled Differential Privacy over Kernel Densities (ICLR 2025 · 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.