Research map / Federated learning and privacy / Differential privacy
Fast Private Kernel Density Estimation via Locality Sensitive Quantization
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 LSQ-RFF instead of prior Gaussian DP-KDE mechanisms | Gaussian DP-KDE on high-dimensional data · Covertype; GloVe | KDE approximation error | Helped |
| Use LSQ-FGT instead of LSQ-RFF | Gaussian DP-KDE on low-dimensional data · Diabetes; NYC Taxi | KDE approximation error | Helped |
| Use LSQ-based mechanisms instead of NoisySample and Bernstein | DP-KDE · Diabetes; NYC Taxi | KDE approximation error | Helped |
| Increase the number of Fourier features beyond the error-optimal setting (ablation) | LSQ-RFF · Gaussian DP-KDE | KDE approximation error | Hurt |
| Increase the FGT truncation parameter beyond its error-optimal setting (ablation) | LSQ-FGT · Gaussian DP-KDE | KDE approximation error | Hurt |
| Elide Laplace noise addition (ablation) | Non-private RFF and FGT variants · Gaussian DP-KDE | KDE approximation error | Helped |
| Add coordinate-specific, refined sensitivity bounds for FGT noise (ablation) | LSQ-FGT · Gaussian DP-KDE | KDE approximation error | Helped |
| Reduce the RFF feature count as privacy becomes stronger (ablation) | LSQ-RFF · Gaussian DP-KDE | KDE approximation error | Mixed |
| Increase the number of Fourier features (ablation) | Non-private RFF · Gaussian DP-KDE | KDE approximation error | Helped |
Closest papers by meaning
- On Differential Private $\ell_1$, $\ell_2$ and $\ell_p^p$ Distance Queries (NeurIPS 2026 · similarity 0.88)
- Differential Privacy in Scalable General Kernel Learning via $K$-means Nystr{\"o}m Random Features (NeurIPS 2024 · similarity 0.88)
- Locality Sensitive Hashing for p-Exponential Kernels with Applications to Density Estimation (NeurIPS 2026 · similarity 0.88)
- SD-KDE: Score-Debiased Kernel Density Estimation (NeurIPS 2025 · similarity 0.85)
- Learning from End User Data with Shuffled Differential Privacy over Kernel Densities (ICLR 2025 · similarity 0.84)
- Efficiently Computing Similarities to Private Datasets (ICLR 2024 · similarity 0.84)
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.