Differentially Private Optimization for Smooth Nonconvex ERM
Not on the map. Why?
Abstract (arXiv)
We develop simple differentially private optimization algorithms that move along directions of (expected) descent to find an approximate second-order solution for nonconvex ERM. We use line search, mini-batching, and a two-phase strategy to improve the speed and practicality of the algorithm. Numerical experiments demonstrate the effectiveness of these approaches.
Key facts. Cited by 5 papers in the map (data through 7 Oct 2026). Builds on 8 of its 18 references that are in the map. First posted on arXiv 9 Feb 2023.
Builds on
8 of 18 references.
References · 8
- 2016Concentrated Differential Privacy: Simplifications, Extensions, and Lower Boundscited by 339
- 2018Privacy Amplification by Subsampling: Tight Analyses via Couplings and Divergencescited by 237
- 2018Subsampled R'enyi Differential Privacy and Analytical Moments Accountantcited by 234
- 2018Differentially Private Empirical Risk Minimization Revisited: Faster and More Generalcited by 127
- 2017Sub-sampled Cubic Regularization for Non-convex Optimizationcited by 63
- 2017Efficient Private ERM for Smooth Objectivescited by 61
- 2016Accelerated Methods for Non-Convex Optimizationcited by 20
- 2020Stochastic Adaptive Line Search for Differentially Private Optimizationcited by 9
Built on by
Citing papers per year: 2023: 2; 2024: 3.
5 papers.
- 2023Breaking the Trilemma of Privacy, Utility, Efficiency via Controllable Machine Unlearningcited by 21
- 2023Private (Stochastic) Non-Convex Optimization Revisited: Second-Order Stationary Points and Excess Riskscited by 12
- 2024How to Make the Gradients Small Privately: Improved Rates for Differentially Private Non-Convex Optimizationcited by 10
- 2024Private Heterogeneous Federated Learning Without a Trusted Server Revisited: Error-Optimal and Communication-Efficient Algorithms for Convex Lossescited by 8
- 2024Adaptive Batch Size for Privately Finding Second-Order Stationary Pointscited by 2
- 2024Adaptive Batch Size for Privately Finding Second-Order Stationary Points10 Oct 2024
- 2024Private Heterogeneous Federated Learning Without a Trusted Server Revisited: Error-Optimal and Communication-Efficient Algorithms for Convex Losses12 Jul 2024
- 2024How to Make the Gradients Small Privately: Improved Rates for Differentially Private Non-Convex Optimization17 Feb 2024
- 2023Breaking the Trilemma of Privacy, Utility, Efficiency via Controllable Machine Unlearning28 Oct 2023
- 2023Private (Stochastic) Non-Convex Optimization Revisited: Second-Order Stationary Points and Excess Risks20 Feb 2023