Concentrated Differential Privacy: Simplifications, Extensions, and Lower Bounds

2016/05/06 by Mark Bun, Thomas Steinke, Bun, Mark +1 · 41 citations
Computer Science · Social Sciences · #Cryptography and Data Security #Cryptography and Security (cs.CR) #Data Structures and Algorithms (cs.DS) #FOS: Computer and information sciences #Information Theory (cs.IT) #Machine Learning (cs.LG) #Privacy, Security, and Data Protection #Privacy-Preserving Technologies in Data

paper · pdf · doi:10.48550/arxiv.1605.02065

openalex publication_date 2016/05/06 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

"Concentrated differential privacy" was recently introduced by Dwork and Rothblum as a relaxation of differential privacy, which permits sharper analyses of many privacy-preserving computations. We present an alternative formulation of the concept of concentrated differential privacy in terms of the Renyi divergence between the distributions obtained by running an algorithm on neighboring inputs. With this reformulation in hand, we prove sharper quantitative results, establish lower bounds, and raise a few new questions. We also unify this approach with approximate differential privacy by giving an appropriate definition of "approximate concentrated differential privacy."

Cited by

• Rethinking Knowledge Distillation in Collaborative Machine Learning: Memory, Knowledge, and Their Interactions
• Privacy Loss of Noise Perturbation via Concentration Analysis of A Product Measure
• "Having Confidence in My Confidence Intervals": How Data Users Engage with Privacy-Protected Wikipedia Data
• When Privacy Isn't Synthetic: Hidden Data Leakage in Generative AI Models
• Differentially Private and Federated Structure Learning in Bayesian Networks
• Sliced Rényi Pufferfish Privacy: Directional Additive Noise Mechanism and Private Learning with Gradient Clipping
• Infinitely divisible privacy and beyond I: resolution of the s²=2k conjecture
• Beyond Membership: Limitations of Add/Remove Adjacency in Differential Privacy
• Differentially private testing for relevant dependencies in high dimensions
• Rényi Differential Privacy for Heavy-Tailed SDEs via Fractional Poincaré Inequalities
• On the Gradient Complexity of Private Optimization with Private Oracles
• Beyond One-Size-Fits-All: Neural Networks for Differentially Private Tabular Data Synthesis
• Nearly-Optimal Private Selection via Gaussian Mechanism
• Differentially Private In-Context Learning with Nearest Neighbor Search
• Mitigating Privacy-Utility Trade-off in Decentralized Federated Learning via f-Differential Privacy
• How to Get Actual Privacy and Utility from Privacy Models: the k-Anonymity and Differential Privacy Families
• An information theorist's tour of differential privacy
• Private Federated Multiclass Post-hoc Calibration
• Differentially Private Synthetic Text Generation for Retrieval-Augmented Generation (RAG)
• The Five Safes as a Privacy Context
• Differentially Private Learning of Exponential Distributions: Simple Algorithms and Tight Bounds
• Privately Estimating Black-Box Statistics
• Differentially Private Two-Stage Gradient Descent for Instrumental Variable Regression
• Accuracy-First Rényi Differential Privacy and Post-Processing Immunity
• Efficiently Attacking Memorization Scores
• Practitioners' Perspectives on a Differential Privacy Deployment Registry
• Rao Differential Privacy
• Public Data Assisted Differentially Private In-Context Learning
• An Interactive Framework for Finding the Optimal Trade-off in Differential Privacy
• Beyond Ordinary Lipschitz Constraints: Differentially Private Stochastic Optimization with Tsybakov Noise Condition
• Statistics-Friendly Confidentiality Protection for Establishment Data, with Applications to the QCEW
• A Tight Context-aware Privacy Bound for Histogram Publication
• Enhancing Differentially Private Linear Regression via Public Second-Moment
• Private Hyperparameter Tuning with Ex-Post Guarantee
• Towards Reliable and Generalizable Differentially Private Machine Learning (Extended Version)
• Mo' Memory, Mo' Problems: Stream-Native Machine Unlearning
• Multi-Hop Privacy Propagation for Differentially Private Federated Learning in Social Networks
• Enhancing Privacy in Decentralized Min-Max Optimization: A Differentially Private Approach
• FedMeNF: Privacy-Preserving Federated Meta-Learning for Neural Fields
• Decentralized Differentially Private Power Method
• Towards Privacy-preserving Photorealistic Self-avatars in Mixed Reality

Related

• Concentrated Differential Privacy
• Lower Bounds for Rényi Differential Privacy in a Black-Box Setting
• A Better Bound Gives a Hundred Rounds: Enhanced Privacy Guarantees via f-Divergences
• Numerical Composition of Differential Privacy
• Three Variants of Differential Privacy: Lossless Conversion and\n Applications