GUPT: privacy preserving data analysis made easy

By Prashanth Mohan, Abhradeep Thakurta, Elaine Shi, Dawn Song, David E. Culler. 2012

267 citations recorded by the source corpus.

Subfields: Privacy & Security

Citations in the corpus

• Detecting Violations of Differential Privacy (2018, neutral): “a while preserving privacy. There are two main approaches to tackling this prevalence of bugs: programming platforms and verification. Programming platforms, such as PINQ [29], Airavat [37], and GUPT [32] provide a small set of primitive operations that can be used as building blocks of algorithms f”
• Prochlo: Strong Privacy for Analytics in the Crowd (2017, neutral): “The analyzer, which has the most complete view of the reported data, may choose from a wide variety of differentially private mechanisms and data processing systems [27,31,38, 53, 63].”
• Privately Learning High-Dimensional Distributions (2018, unknown): “Nissim et al. (2007); Dwork and Lei (2009)) that make it possible to add noise roughly proportional to average sensitivity for univariate statistics.”
• K-variates++: More Pluses in the K-means++ (2016, unknown): “In comparison, GUPT adds noise ∝ kR/(` ) at the end of this aggregation process.”
• Combining Differential Privacy and Secure Multiparty Computation (2015, unknown): “We report on the experience that we have obtained with the implementations of GUPT’s sample-and-aggregate method [31] and the provenance for PDP method [14] on top of the Sharemind SMC framework.”
• PROCHLO: Strong Privacy for Analytics in the Crowd (2017, unknown): “The analyzer, which has the most complete view of the reported data, may choose from a wide variety of differentially private mechanisms and data processing systems [27,31,38, 53, 63].”
• Proving differential privacy with shadow execution (2019, unknown): “rivacy is critical to its success. The initial line of work on formal verification for differential privacy (e.g., [6–10]) was concerned with increasing expressiveness. A parallel line of work (e.g., [33, 35, 37, 44]) focuses more on usability – on developing platforms that keep track of the privacy”
• PrivateClean: Data Cleaning and Differential Privacy (2016, unknown): “The GUPT project explored the problem of (a “privacy budget”) given desired accuracy guarantees [31].”
• AutoPriv: Automating Differential Privacy Proofs (2016, unknown): “jjh ala (2015) and Lyu et al. (2016) catalog some of the most recent cases. Currently, there are two strategies for combatting this trend. The first is the use of programming platforms (McSherry 2009; Mohan et al. 2012; Roy et al. 2010) that have privacy primitives that restrict the privacy-preservin”
• Low Latency Geo-distributed Data Analytics (2015, unknown): “Additionally, regulatory and privacy concerns might also forbid central aggregation [42].”
• Differentially Private Data Publishing and Analysis: A Survey (2017, unknown): “In general, even some differentially private platforms such as SuLQ [53], PINQ [12], Privgene [59], Gupt [60] can automatically implement clustering algorithms.”
• PrivBayes: private data release via bayesian networks (2014, unknown): “Even if the data is skewed in the domain space, i.e., there are some entries x[i] with high counts, such peaks are infrequent and so the vast majority of published values x∗[i] are useless.”
• Differentially Private K-Means Clustering (2015, unknown): “The k-means clustering problem was also used to motivate the sample and aggregate framework (SAF) for satisfying differential privacy, which was developed in [32, 38], and implemented in the GUPT system [31].”
• LightDP: towards automating differential privacy proofs (2016, unknown): “The first is the use of programming platforms (McSherry 2009; Mohan et al. 2012; Roy et al. 2010) that have privacy primitives that restrict the privacy-preserving algorithms that can be implemented and often add more noise than is necessary to the computation.”