Research map / Federated learning and privacy / Differential privacy
Differentially Private Empirical Risk Minimization with Non-convex Loss Functions
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 random output sample w_j (Theorem 3) instead of final iterate w_T for DP-GLD | DP-GLD (Algorithm 1) · theoretical bound on empirical risk | excess empirical risk bound (n-dependence) | Helped |
| Use DP-FW-L1 (Frank-Wolfe with exponential mechanism over vertices of L1 ball) for GLM and robust regression under L_inf data / L1 constraint | GLM and robust regression · theoretical population risk bound | population risk bound dependence on dimension | Helped |
| Add Gaussian-noise-based escape from saddle points (DP-GD, Algorithm 4) to find approximate second-order stationary points privately | DP gradient descent for non-convex ERM · theoretical sample complexity for alpha-SOSP | required sample size n | No clear effect |
| Add a selection step (Algorithm 5) that privately tests candidate iterates for gradient and Hessian condition | DP-GD (Algorithm 4) · theoretical sample complexity for alpha-SOSP | required sample size n | Hurt |
Closest papers by meaning
- Faster Algorithms for User-Level Private Stochastic Convex Optimization (NeurIPS 2024 · similarity 0.90)
- Differentially Private Optimization with Sparse Gradients (NeurIPS 2024 · similarity 0.90)
- Adapting to Linear Separable Subsets with Large-Margin in Differentially Private Learning (ICML 2025 · similarity 0.89)
- Differential Private Stochastic Optimization with Heavy-tailed Data: Towards Optimal Rates (AAAI 2025 · similarity 0.89)
- Improved Rates of Differentially Private Nonconvex-Strongly-Concave Minimax Optimization (AAAI 2025 · similarity 0.89)
- Improved Sample Complexity for Private Nonsmooth Nonconvex Optimization (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.