E-differential Privacy Research Articles

A data- and workload-aware algorithm for range queries under differential privacy

We describe a new algorithm for answering a given set of range queries under ε-differential privacy which often achieves substantially lower error than competing methods. Our algorithm satisfies differential privacy by adding noise that is adapted to the input data and to the given query set. We first privately learn a partitioning of the domain into buckets that suit the input data well. Then we privately estimate counts for each bucket, doing so in a manner well-suited for the given query set. Since the performance of the algorithm depends on the input database, we evaluate it on a wide range of real datasets, showing that we can achieve the benefits of data-dependence on both "easy" and "hard" databases.

Optimal data-independent noise for differential privacy

ε-Differential privacy is a property that seeks to characterize privacy in data sets. It is formulated as a query-response method, and computationally achieved by output perturbation. Several noise-addition methods to implement such output perturbation have been proposed in the literature. We focus on data-independent noise, that is, noise whose distribution is constant across data sets. Our goal is to find the optimal data-independent noise distribution to achieve ε-differential privacy. We propose a general optimality criterion based on the concentration of the probability mass of the noise distribution around zero, and we show that any noise optimal under this criterion must be optimal under any other sensible criterion. We also show that the Laplace distribution, commonly used for noise in ε-differential privacy, is not optimal, and we build the optimal data-independent noise distribution. We compare the Laplace and the optimal data-independent noise distributions. For univariate query functions, both introduce a similar level of distortion; for multivariate query functions, optimal data-independent noise offers responses with substantially better data quality.

Practical differential privacy via grouping and smoothing

We address one-time publishing of non-overlapping counts with ε-differential privacy. These statistics are useful in a wide and important range of applications, including transactional, traffic and medical data analysis. Prior work on the topic publishes such statistics with prohibitively low utility in several practical scenarios. Towards this end, we present GS, a method that pre-processes the counts by elaborately grouping and smoothing them via averaging. This step acts as a form of preliminary perturbation that diminishes sensitivity, and enables GS to achieve ε-differential privacy through low Laplace noise injection. The grouping strategy is dictated by a sampling mechanism, which minimizes the smoothing perturbation. We demonstrate the superiority of GS over its competitors, and confirm its practicality, via extensive experiments on real datasets.

Privacy-preserving heterogeneous health data sharing

Privacy-preserving data publishing addresses the problem of disclosing sensitive data when mining for useful information. Among existing privacy models, ε-differential privacy provides one of the strongest privacy guarantees and makes no assumptions about an adversary's background knowledge. All existing solutions that ensure ε-differential privacy handle the problem of disclosing relational and set-valued data in a privacy-preserving manner separately. In this paper, we propose an algorithm that considers both relational and set-valued data in differentially private disclosure of healthcare data. The proposed approach makes a simple yet fundamental switch in differentially private algorithm design: instead of listing all possible records (ie, a contingency table) for noise addition, records are generalized before noise addition. The algorithm first generalizes the raw data in a probabilistic way, and then adds noise to guarantee ε-differential privacy. We showed that the disclosed data could be used effectively to build a decision tree induction classifier. Experimental results demonstrated that the proposed algorithm is scalable and performs better than existing solutions for classification analysis. The resulting utility may degrade when the output domain size is very large, making it potentially inappropriate to generate synthetic data for large health databases. Unlike existing techniques, the proposed algorithm allows the disclosure of health data containing both relational and set-valued data in a differentially private manner, and can retain essential information for discriminative analysis.

Differential Privacy and Statistical Disclosure Risk Measures

We compare the disclosure risk criterion of e-differential privacy with a criterion based on probabilities that intruders uncover actual values given the released data. To do so, we generate fully ...

Functional mechanism

ε-differential privacyis the state-of-the-art model for releasing sensitive information while protecting privacy. Numerous methods have been proposed to enforce ε-differential privacy in various analytical tasks, e.g.,regression analysis. Existing solutions for regression analysis, however, are either limited to non-standard types of regression or unable to produce accurate regression results. Motivated by this, we propose theFunctional Mechanism, a differentially private method designed for a large class of optimization-based analyses. The main idea is to enforce ε-differential privacy by perturbing theobjective functionof the optimization problem, rather than its results. As case studies, we apply the functional mechanism to address two most widely used regression models, namely,linear regressionandlogistic regression. Both theoretical analysis and thorough experimental evaluations show that the functional mechanism is highly effective and efficient, and it significantly outperforms existing solutions.

Publishing Search Logs—A Comparative Study of Privacy Guarantees

Search engine companies collect the “database of intentions,” the histories of their users' search queries. These search logs are a gold mine for researchers. Search engine companies, however, are wary of publishing search logs in order not to disclose sensitive information. In this paper, we analyze algorithms for publishing frequent keywords, queries, and clicks of a search log. We first show how methods that achieve variants of k-anonymity are vulnerable to active attacks. We then demonstrate that the stronger guarantee ensured by ε-differential privacy unfortunately does not provide any utility for this problem. We then propose an algorithm ZEALOUS and show how to set its parameters to achieve (ε, δ)-probabilistic privacy. We also contrast our analysis of ZEALOUS with an analysis by Korolova et al. [17] that achieves (ε',δ')-indistinguishability. Our paper concludes with a large experimental study using real applications where we compare ZEALOUS and previous work that achieves k-anonymity in search log publishing. Our results show that ZEALOUS yields comparable utility to k-anonymity while at the same time achieving much stronger privacy guarantees.

Differentially Private Empirical Risk Minimization.

Privacy-preserving machine learning algorithms are crucial for the increasingly common setting in which personal data, such as medical or financial records, are analyzed. We provide general techniques to produce privacy-preserving approximations of classifiers learned via (regularized) empirical risk minimization (ERM). These algorithms are private under the ε-differential privacy definition due to Dwork et al. (2006). First we apply the output perturbation ideas of Dwork et al. (2006), to ERM classification. Then we propose a new method, objective perturbation, for privacy-preserving machine learning algorithm design. This method entails perturbing the objective function before optimizing over classifiers. If the loss and regularizer satisfy certain convexity and differentiability criteria, we prove theoretical results showing that our algorithms preserve privacy, and provide generalization bounds for linear and nonlinear kernels. We further present a privacy-preserving technique for tuning the parameters in general machine learning algorithms, thereby providing end-to-end privacy guarantees for the training process. We apply these results to produce privacy-preserving analogues of regularized logistic regression and support vector machines. We obtain encouraging results from evaluating their performance on real demographic and benchmark data sets. Our results show that both theoretically and empirically, objective perturbation is superior to the previous state-of-the-art, output perturbation, in managing the inherent tradeoff between privacy and learning performance.