Practical privacy: the SuLQ framework

By Avrim Blum, Cynthia Dwork, Frank McSherry, Kobbi Nissim. 2005

982 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Theory and Mathematics, Privacy & Security

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 Computational Theory and Mathematics

In Privacy & Security

Citations in the corpus

Federated Learning With Differential Privacy: Algorithms and Performance Analysis (2019, background): “A natural approach to preventing information leakage is to add artificial noise, one prominent example of which is differential privacy (DP) [24], [25].”
IBM Federated Learning: an Enterprise Framework White Paper V0.1 (2020, methodology): “Techniques such as local diﬀerential privacy [8], multi-party computation (SMC) [13, 33] and hybrid approaches have been proposed for this purpose [31].”
Private Counting from Anonymous Messages: Near-Optimal Accuracy with Vanishing Communication Overhead (2020, methodology): “As observed in Blum et al. (2005), binary summation is of particular interest in ML since it is sufficient for implementing any learning algorithm based on statistical queries (Kearns, 1998), which includes most of the known PAC-learning algorithms.”
Tight Lower Bounds for Differentially Private Selection (2017, methodology): “We can ensure differential privacy by adding carefully calibrated noise to the empirical mean X to obtain a noisy empirical mean X̃ [13], [14], [15], [12].”
Prio: Private, Robust, and Scalable Computation of Aggregate Statistics (2017, background): “A variant of the same principle is to have a trusted proxy (as in SuLQ [21] and PDDP [34]) or a set of minimally trusted servers [92] add noise to already-collected data.”
The Composition Theorem for Differential Privacy (2013, background): “, 2005; 2013) and they provide a primitive for constructing more complex queries (Blum et al., 2005).”
Boosting and Differential Privacy (2010, background): “As in previous works on differentially private learning [2], [17], we can also view the input database as a training set in a learning algorithm, where each row corresponds to an element in the training set.”
On the Difficulties of Disclosure Prevention in Statistical Databases or The Case for Differential Privacy (2010, background): “See also [30, 28, 5, 4, 30, 3, 27] for other works dealing with the notion.”
Differentially Private Empirical Risk Minimization: Efficient Algorithms and Tight Error Bounds (2014, background): “Differential privacy is a rigorous notion of privacy that emerged from a line of work in theoretical computer science and cryptography [19, 6, 21].”
Security Evaluation of Support Vector Machines in Adversarial Environments (2014, background): “data [15], measuring the sensitivity of the SVM appears to be non-trivial owing to the non-linear influence an individual training datum may have on the learned SVM.”
Learning in a Large Function Space: Privacy-Preserving Mechanisms for SVM Learning (2009, background): “…interval queries and half-spaces as explored by Blum et al. 2008), or on constructing learning algorithms that can be decomposed into subset-sum operations (e.g., perceptron, k-NN, ID3 as described by Blum et al. 2005, and various recommender systems due to the work of McSherry and Mironov 2009).”
Distributed Private Data Analysis: On Simultaneously Solving How and What (2008, background): “A criterion for functions that preserve privacy of individual entries, differential privacy, has evolved in a sequence of recent works [7, 15, 14, 2, 11, 8, 9].”
On the geometry of differential privacy (2009, neutral): “Private mechanisms with small error, for other classes of queries have also been studied in several other works, see e.g. [BDMN05, BCD + 07, MT07, CM08, GLM + 10].”
A Statistical Framework for Differential Privacy (2008, methodology): “This framework is used in Blum et al. (2008). For the rest of this section, assume that Z = (Z1, .”
Robust Traceability from Trace Amounts (2015, background): “Like the fingerprinting attacks of [14], [18], the lower bounds on noise implied by our attacks nearly match the upper bounds on noise sufficient to ensure the strong guarantees of differential privacy, for example, via the Gaussian or Laplace mechanisms [3], [21]–[24]).”
A learning theory approach to non-interactive database privacy (2008, neutral): “Categories and Subject Descriptors F.2 [Theory of Computation]: Analysis of Algorithms and Problem Complexity General Terms Algorithms, Security, Theory Keywords non-interactive database privacy, learning theory”