Research map / Optimization / Nonconvex and smooth optimization
Differential Privacy Dynamics of Langevin Diffusion and Noisy Gradient Descent
Rishav Chourasia, Jiayuan Ye, Reza Shokri
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 the privacy-dynamics RDP bound instead of composition-based privacy accounting | Noisy GD · Noisy GD training over K iterations | RDP privacy loss: composition-based bound grows linearly with K; proposed bound is proportional to (1-e^{-\lambda\etaK/2}) | Improved |
| Require smoothness and strong convexity, with an initialization that satisfies the stated Gaussian condition | Noisy GD privacy guarantee · Noisy GD on neighboring datasets | RDP privacy loss: \varepsilon = \frac{\alpha S_g^2}{\lambda \sigma^2n^2}(1-e^{-\lambda\etaK/2}) | Improved |
| Apply the analysis to squared-norm loss and use its tighter LSI constant | Noisy GD · Squared-norm loss | RDP privacy bound: \frac{\alpha S_g^2 }{(2-\eta)\sigma^2n^2}(1-e^{-\frac{2-\eta}{2}\etaK}) | Improved |
| Use Noisy GD rather than Noisy SGD for the stated strongly convex utility regime | Privacy-preserving ERM · Lipschitz, smooth, strongly convex ERM | gradient complexity: a factor of $n/{\log n}$ | Improved |
Closest papers by meaning
- Shifted Interpolation for Differential Privacy (ICML 2024 · similarity 0.91)
- Convergent Privacy Loss of Noisy-SGD without Convexity and Smoothness (ICLR 2025 · similarity 0.89)
- Certification for Differentially Private Prediction in Gradient-Based Training (ICML 2025 · similarity 0.88)
- Differentially Private Stochastic Gradient Descent with Fixed-Size Minibatches: Tighter RDP Guarantees with or without Replacement (NeurIPS 2024 · similarity 0.88)
- Privacy without Noisy Gradients: Slicing Mechanism for Generative Model Training (NeurIPS 2024 · similarity 0.88)
- Differential Privacy Guarantees of Markov Chain Monte Carlo Algorithms (ICML 2025 · similarity 0.88)
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.
