Histogramming Privately Ever After: Differentially-Private Data-Dependent Error Bound Optimisation

Fanaeepour, Maryam; Rubinstein, Benjamin I. P.

doi:10.1109/icde.2018.00111

Cited by 9 publications

(5 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the first step, k clusters are randomly initialized by choosing k centroids and arranging each record to its nearest centroid (Line 5). In the second step, each record's label is reset to the closest cluster in the previous iteration (Lines [6][7][8][9][10][11][12][13].…”

Section: Clustering On Anonymized Datamentioning

confidence: 99%

See 1 more Smart Citation

PrivBV: Distance-aware encoding for distributed data with local differential privacy

Sun

Ping

Yao

2022

Tsinghua Sci. Technol.

View full text Add to dashboard Cite

Recently, local differential privacy (LDP) has been used as the de facto standard for data sharing and analyzing with high-level privacy guarantees. Existing LDP-based mechanisms mainly focus on learning statistical information about the entire population from sensitive data. For the first time in the literature, we use LDP for distance estimation between distributed data to support more complicated data analysis. Specifically, we propose PrivBV-a locally differentially private bit vector mechanism with a distance-aware property in the anonymized space. We also present an optimization strategy for reducing privacy leakage in the high-dimensional space. The distance-aware property of PrivBV brings new insights into complicated data analysis in distributed environments.As study cases, we show the feasibility of applying PrivBV to privacy-preserving record linkage and non-interactive clustering. Theoretical analysis and experimental results demonstrate the effectiveness of the proposed scheme.

show abstract

Section: Clustering On Anonymized Datamentioning

confidence: 99%

“…The -LDP provides high privacy-preserving levels by guaranteeing that the probability of any two different inputs being projected into the same output is bounded by e . LDP has shown great significance in learning statistical information, especially for the mean and frequency estimations (heavy hitter estimation) [8][9][10][11] .…”

Section: Introductionmentioning

confidence: 99%

PrivBV: Distance-aware encoding for distributed data with local differential privacy

Sun

Ping

Yao

2022

Tsinghua Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…Hamburger, 0.8 , Fries, −0.3 , Pepsi, 0.9 user 3 Salad, 0.75 , Fries, −0.1 , Pepsi, 0.8 ... ... potential adversaries (including the aggregator in LDP). The local differential privacy has been widely used in crowdsourcing and IoT scenarios for privacy-preserving data analytics [4], [5]. To analyze user's data with high-level privacy guarantees, respected data service providers have applied local differential in their services.…”

Section: Usersmentioning

confidence: 99%

Conditional Analysis for Key-Value Data with Local Differential Privacy

Sun,

Zhao,

et al. 2019

Preprint

View full text Add to dashboard Cite

Local differential privacy (LDP) has been deemed as the de facto measure for privacy-preserving distributed data collection and analysis. Recently, researchers have extended LDP to the basic data type in NoSQL systems: the key-value data, and show its feasibilities in mean estimation and frequency estimation. In this paper, we develop a set of new perturbation mechanisms for key-value data collection and analysis under the strong model of local differential privacy. Since many modern machine learning tasks rely on the availability of conditional probability or the marginal statistics, we then propose the conditional frequency estimation method for key analysis and the conditional mean estimation for value analysis in key-value data. The released statistics with conditions can further be used in learning tasks. Extensive experiments of frequency and mean estimation on both synthetic and real-world datasets validate the effectiveness and accuracy of the proposed key-value perturbation mechanisms against the state-of-art competitors.

show abstract

“…However, values of that are deemed too high can subject a company to criticisms of not providing enough privacy [42]. For this reason, researchers invest significant effort in tuning algorithms [11,47,29,1,40,22] and privacy analyses [8,38,40,20] to provide better utility while using smaller privacy budgets.…”

Section: Introductionmentioning

confidence: 99%

Free Gap Estimates from the Exponential Mechanism, Sparse Vector, Noisy Max and Related Algorithms

Ding¹,

Wang²,

Xiao³

et al. 2020

Preprint

View full text Add to dashboard Cite

Private selection algorithms, such as the Exponential Mechanism, Noisy Max and Sparse Vector, are used to select items (such as queries with large answers) from a set of candidates, while controlling privacy leakage in the underlying data. Such algorithms serve as building blocks for more complex differentially private algorithms. In this paper we show that these algorithms can release additional information related to the gaps between the selected items and the other candidates for free (i.e., at no additional privacy cost). This free gap information can improve the accuracy of certain follow-up counting queries by up to 66%. We obtain these results from a careful privacy analysis of these algorithms. Based on this analysis, we further propose novel hybrid algorithms that can dynamically save additional privacy budget.

show abstract

Histogramming Privately Ever After: Differentially-Private Data-Dependent Error Bound Optimisation

Cited by 9 publications

References 11 publications

PrivBV: Distance-aware encoding for distributed data with local differential privacy

PrivBV: Distance-aware encoding for distributed data with local differential privacy

Conditional Analysis for Key-Value Data with Local Differential Privacy

Free Gap Estimates from the Exponential Mechanism, Sparse Vector, Noisy Max and Related Algorithms

Contact Info

Product

Resources

About