Differential Privacy for Statistics: What we Know and What we Want to Learn

By Cynthia Dwork, Adam D. Smith. 2010

321 citations recorded by the source corpus.

We motivate and review the denition privacy, our principal motivating scenario was a statistical database, in which the trusted and trustworthy curator (in our minds, the Census Bureau) gathers sensitive information from a large number of respondents (the sample), with the goal of learning and releasing to the public statistical facts about the underlying population. The diculty, of course, is to release statistical information without compromising the privacy of the individual respondents. We initially only thought in terms of a noninteractive setting, in which the curator computes and publishes some statistics, and the data are not used further. Privacy concerns may aect

Subfields: Artificial Intelligence, Public Health, Environmental and Occupational Health, Sociology and Political Science

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Public Health, Environmental and Occupational Health

In Sociology and Political Science

Citations in the corpus

DISCO: Dynamic and Invariant Sensitive Channel Obfuscation for deep neural networks (2020, methodology): “Additionally, techniques such as multi-party computation [14, 15] and differential privacy [16, 17, 18, 19] have been employed to improve the privacy in federated-learning.”
Signal Processing and Machine Learning with Differential Privacy (2013, background): “There have been other surveys of differential privacy literature; in particular, Dwork and Smith's survey [8] [ Algorithms and challenges for continuous data ]”
Dependence Makes You Vulnberable: Differential Privacy Under Dependent Tuples (2016, background): “Differential privacy is a rigorous mathematical framework aimed at protecting the privacy of users’ sensitive information in a statistical database [12]–[16].”
The Right Complexity Measure in Locally Private Estimation: It is not the Fisher Information (2018, background): “In response to these challenges, researchers in the statistics, databases, and computer science communities have studied diﬀerential privacy [54, 25, 24, 23, 27, 21] as a formalization of disclosure risk limitation, providing strong privacy guarantees.”
Efficient Use of Differentially Private Binary Trees (2015, background): “Thus the release of a private binary tree can be a broadly useful privacy preserving means of allowing exploration of a variable, as for example in the statistical release of a non-interactive curator (Dwork and Smith, 2009).”
DEEProtect: Enabling Inference-based Access Control on Mobile Sensing Applications (2017, background): “Differential privacy (DP) [24], [27], [25], [29], [26] is a rigorous mathematical framework that prevents an attacker from inferring the presence or absence of a particular record in a statistical database.”
Randomized requantization with local differential privacy (2016, methodology): “In order to understand these tradeoffs, we adopt the differential privacy framework [1, 2] for quantifying privacy risk.”
Utility-Privacy Tradeoffs in Databases: An Information-Theoretic Approach (2011, background): “base adds noise to the response based on a data model. Differential privacy: More recently, privacy approaches for statistical databases have been driven by the ǫ-differential privacy deﬁnition [24], [36]–[38] where ǫ is a parameter that determines the privacy threshold (smaller ǫ implies tighter pr”
Optimal data-independent noise for differential privacy (2013, methodology): “For a more complete overview of differential privacy and, in particular, of a variety of methods used to attain it, see [7,8,12].”
Differential Privacy and Machine Learning: a Survey and Review (2014, background): “Several recent surveys address differential privacy and data science [17, 47, 22].”
Per-instance Differential Privacy and the Adaptivity of Posterior Sampling in Linear and Ridge regression (2017, methodology): “Meanwhile, there are algorithms that attain asymptotic eﬃciency either by subsample-and-aggregate [28] or by simply adding noise to the suﬃcient statistics [14, 17].”
Differentially Private Bayesian Linear Regression (2019, background): “…(OPS) [Dimitrakakis et al., 2014, Geumlek et al., 2017, Minami et al., 2016, Wang, 2018, Wang et al., 2015, Zhang et al., 2016], subsample and aggregate [Dwork and Smith, 2010, Smith, 2008], objective perturbation [Kifer et al., 2012], and noisy stochastic gradient descent [Bassily et al., 2014].”
Differential Privacy for Clinical Trial Data: Preliminary Evaluations (2009, background): “Following [11], [10], and [16], we say that two data setsx = (x1, x2, .”
Bayesian Differential Privacy through Posterior Sampling (2013, background): “In this case privacy was still achieved by performing a differentially-private test on local sensitivity before release (Dwork and Smith, 2009).”
Sample Complexity Bounds on Differentially Private Learning via Communication Complexity (2014, background): “…sensitive, for example in the case of medical or ﬁnancial information, and therefore privacy-preserving data analysis is becoming an increasingly important area of study in machine learning, data mining and statistics [Dwork and Smith, 2009, Sarwate and Chaudhuri, 2013, Dwork and Roth, 2014].”