Abstract: We study statistical risk minimization problems under a privacy model in which the data is kept confidential even from the learner. In this local privacy framework, we establish sharp upper and lower bounds on the convergence rates of statistical estimation procedures. As a consequence, we exhibit a precise tradeoff between the amount of privacy the data preserves and the utility, as measured by convergence rate, of any statistical estimator or learning procedure.
Recommendations
- Minimax Optimal Procedures for Locally Private Estimation
- What can we learn privately?
- The cost of privacy: optimal rates of convergence for parameter estimation with differential privacy
- Geometrizing rates of convergence under local differential privacy constraints
- Differentially private empirical risk minimization
Cites work
- A statistical framework for differential privacy
- Acceleration of Stochastic Approximation by Averaging
- Asymptotic Statistics
- Compressed and Privacy-Sensitive Sparse Regression
- Convergence of estimates under dimensionality restrictions
- Differential privacy and robust statistics
- Differential privacy for functions and functional data
- Differential Privacy: A Survey of Results
- Differentially private empirical risk minimization
- Efficient noise-tolerant learning from statistical queries
- Elements of Information Theory
- Estimating Risks of Identification Disclosure in Microdata
- Foundations of Modern Probability
- scientific article; zbMATH DE number 3123545 (Why is no real title available?)
- scientific article; zbMATH DE number 3908323 (Why is no real title available?)
- scientific article; zbMATH DE number 4060392 (Why is no real title available?)
- scientific article; zbMATH DE number 48436 (Why is no real title available?)
- scientific article; zbMATH DE number 51132 (Why is no real title available?)
- scientific article; zbMATH DE number 2107836 (Why is no real title available?)
- scientific article; zbMATH DE number 5485574 (Why is no real title available?)
- Information theoretic security
- Information-theoretic determination of minimax rates of convergence
- Information-Theoretic Lower Bounds on the Oracle Complexity of Stochastic Convex Optimization
- Lectures on Choquet's theorem
- Mirror descent and nonlinear projected subgradient methods for convex optimization.
- On the geometry of differential privacy
- On the Question of Statistical Confidentiality
- Privacy-preserving statistical estimation with optimal convergence rates
- Randomized response: a survey technique for eliminating evasive answer bias
- Robust Stochastic Approximation Approach to Stochastic Programming
- Theory of Cryptography
- Uniqueness of solution in linear programming
- Universally utility-maximizing privacy mechanisms
- What can we learn privately?
Cited in
(25)- A compressive privacy approach to generalized information bottleneck and privacy funnel problems
- On density estimation at a fixed point under local differential privacy
- Strongly universally consistent nonparametric regression and classification with privatised data
- Distributionally-robust machine learning using locally differentially-private data
- Geometrizing rates of convergence under local differential privacy constraints
- Locally differentially private item-based collaborative filtering
- Optimal locally private estimation under \(\ell_p\) loss for \(1\le p\le 2\)
- Inferring ground truth from crowdsourced data under local attribute differential privacy
- Compressive statistical learning with random feature moments
- Learning with differential privacy: stability, learnability and the sufficiency and necessity of ERM principle
- What can we learn privately?
- The complexity of differential privacy
- Private sequential learning
- Perturbation of convex risk minimization and its application in differential private learning algorithms
- Model averaging with privacy-preserving
- Econometrics with privacy preservation
- scientific article; zbMATH DE number 7415075 (Why is no real title available?)
- Semi-supervised learning with summary statistics
- Differentially private empirical risk minimization
- Privacy-preserving set-based estimation using partially homomorphic encryption
- The right complexity measure in locally private estimation: it is not the Fisher information
- Differentially private reinforcement learning
- Locally covert learning
- Federated Offline Reinforcement Learning
- Optimal learning policies for differential privacy in multi-armed bandits
This page was built for publication: Privacy aware learning
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5501941)
