Research map / Federated learning and privacy / Differential privacy
Benefits and Pitfalls of the Exponential Mechanism with Applications to Hilbert Spaces and Functional PCA
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 |
|---|---|---|---|
| Increase the privacy budget from 1/8 to 2 | Private FPCA · Berkeley growth, k=1 | variance ratio: 0.264 (.024) -> 0.743 (.018) | Helped |
| Increase the privacy budget from 1/8 to 2 | Private FPCA · Berkeley growth, k=2 | variance ratio: 0.494 (.023) -> 0.787 (.012) | Helped |
| Increase the privacy budget from 1/8 to 2 | Private FPCA · Berkeley growth, k=3 | variance ratio: 0.672 (.020) -> 0.855 (.010) | Helped |
| Increase the privacy budget from 1/8 to 2 | Private FPCA · DTI CCA, k=1 | variance ratio: 0.372 (.025) -> 0.933 (.006) | Helped |
| Increase the privacy budget from 1/8 to 2 | Private FPCA · DTI CCA, k=2 | variance ratio: 0.569 (.024) -> 0.928 (.004) | Helped |
| Increase the privacy budget from 1/8 to 2 | Private FPCA · DTI CCA, k=3 | variance ratio: 0.727 (.018) -> 0.939 (.003) | Helped |
| Increase the privacy budget from 1/8 to 2 | Private FPCA · Berkeley growth, k=1 | subspace norm: 0.776 (.025) -> 0.275 (.020) | Helped |
| Increase the privacy budget from 1/8 to 2 | Private FPCA · DTI CCA, k=1 | subspace norm: 0.679 (.026) -> 0.073 (.006) | Helped |
| Release more components, from k=1 to k=3 | Private FPCA · Berkeley growth, privacy budget 1/8 | variance ratio: 0.264 (.024) -> 0.672 (.020) | Helped |
| Release more components, from k=1 to k=3 | Private FPCA · DTI CCA, privacy budget 1/8 | variance ratio: 0.372 (.025) -> 0.727 (.018) | Helped |
| Use the private FPCA exponential mechanism | Functional PCA · Berkeley growth and DTI CCA | variance ratio | Mixed |
Closest papers by meaning
- Differential Privacy without Sensitivity (NeurIPS 2016 · similarity 0.91)
- Optimal Differential Privacy Composition for Exponential Mechanisms (ICML 2020 · similarity 0.90)
- A Joint Exponential Mechanism For Differentially Private Top-$k$ (ICML 2022 · similarity 0.90)
- Exact Privacy Guarantees for Markov Chain Implementations of the Exponential Mechanism with Artificial Atoms (NeurIPS 2021 · similarity 0.89)
- Approximate Differential Privacy of the $\ell_2$ Mechanism (ICML 2025 · similarity 0.88)
- Auditing $f$-differential privacy in one run (ICML 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.