Abstract:Local Differential Privacy (LDP) can provide each user with strong privacy guarantees under untrusted data curators while ensuring accurate statistics derived from privatized data. Due to its powerfulness, LDP has been widely adopted to protect privacy in various tasks (e.g., heavy hitters discovery, probability estimation) and systems (e.g., Google Chrome, Apple iOS). In particular, <math>(∊,δ)</math>-LDP has been studied in related statistical tasks like private learning and hypothesis testing, but is mainly achieved by using Gaussian mechanism, leading to the limited data utility. In this paper, we investigate several novel mechanisms that achieve <math>(∊,δ)</math>-LDP with higher data utility in collecting and analyzing users' data. Specifically, we first design two <math>(∊,δ)</math>-LDP algorithms for mean estimations on multi-dimensional numeric data, which can ensure higher accuracy than the optimal Gaussian mechanism. Then, we investigate different local protocols for frequency estimations on categorical attributes under <math>(∊,δ)</math>-LDP. Based on the proposed mechanisms, we further study on <math>(∊,δ)</math>-LDP-compliant stochastic gradient descent algorithms for machine learning models. Besides, the theoretical analysis of the error bound and the variance of the proposed algorithms are also presented in the paper. We have conducted extensive experiments on both real-world and synthetic datasets and demonstrated the high data utility of our proposed algorithms in the perspectives of simple data statistics tasks and complex machine learning tasks. The experimental results have shown that our proposed algorithms can effectively improve the data utility in different tasks while alleviating the privacy concerns of each individual.

A Low Cost and Un-Cancelled Laplace Noise Based Differential Privacy Algorithm for Spatial Decompositions

On Learning Cluster Coefficient of Private Networks

Differentially Private Spatial Decompositions

Count on Your Elders: Laplace vs Gaussian Noise

Numerical Estimation of Spatial Distributions under Differential Privacy

Differentially private publication for related POI discovery

A novel self-adaptive grid-partitioning noise optimization algorithm based on differential privacy

Distributed Least-Squares Optimization Solvers with Differential Privacy

Infinitely Divisible Noise in the Low Privacy Regime

Conducting Correlated Laplace Mechanism For Differential Privacy

Local Differential Privacy for data collection and analysis

Differential Privacy with Higher Utility by Exploiting Coordinate-wise Disparity: Laplace Mechanism Can Beat Gaussian in High Dimensions

A Workload Division Differential Privacy Algorithm to Improve the Accuracy for Linear Computations

Local Differential Private Data Aggregation for Discrete Distribution Estimation

Differential Private Noise Adding Mechanism: Fundamental Theory and its Application.

Customized Local Differential Privacy for Multi-Agent Distributed Optimization

Efficient Noise Generation Protocols for Differentially Private Multiparty Computation

Answering Spatial Density Queries Under Local Differential Privacy

Differential Private Noise Adding Mechanism and Its Application on Consensus Algorithm

More Flexible Differential Privacy: The Application of Piecewise Mixture Distributions in Query Release