When Random Sampling Preserves Privacy

By Kamalika Chaudhuri, Nina Mishra. 2006

116 citations recorded by the source corpus.

Subfields: Artificial Intelligence, 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 Sociology and Political Science

Citations in the corpus

Federated Learning and Differential Privacy: Software tools analysis, the Sherpa.ai FL framework and methodological guidelines for preserving data privacy (2020, background): “Data obfuscation, anonymisation techniques such as blindly trusting artiﬁcial intelligence black box models (i.e. convolutional neural networks) or randomly sampling client model’s data have been proven to be inadequate to preserve privacy [10, 11].”
AdaCliP: Adaptive Clipping for Private SGD (2019, methodology): “Diﬀerential privacy deﬁnition was originally used to provide strong privacy guarantees for database querying and since used in several applications [20, 21, 22].”
Federated Heavy Hitters Discovery with Differential Privacy (2019, background): “The connection between DP, random sampling, and k-anonymity has previously appeared in the literature [13, 36, 28].”
Differentially Private Empirical Risk Minimization (2009, neutral): “(2006b), and has been used since in numerous works on privacy (Chaudhuri and Mishra, 2006; McSherry and Talwar, 2007; Nissim et al., 2007; Barak et al., 2007; Chaudhuri and Monteleoni, 2008; Machanavajjhala et al., 2008).”
Privacy Amplification by Subsampling: Tight Analyses via Couplings and Divergences (2018, methodology): “Privacy ampliﬁcation with Poisson sampling was used in [Chaudhuri and Mishra, 2006, Beimel et al., 2010, Kasiviswanathan et al., 2011, Beimel et al., 2014], which considered loose bounds.”
Composition attacks and auxiliary information in data privacy (2008, methodology): “However, certain recently developed techniques such as sampling [8], instance-based noise addition [29] and data synthesis [25] appear to require relaxations of the deﬁnition.”
Crowd-Blending Privacy (2012, background): “in providing privacy: In [11] the authors quantify the level of the privacy that may be obtained from just random sampling of data (without any further sanitization); in [12] the authors consider a certain type of “sample-and-aggregate” mechanism for achieving differential privacy (but the sampling ”
Wireless Sensor Networks for Healthcare (2010, background): “For example, Chaudhuri and Mishra showed that personal information may be identified even from anonymized and sanitized population level data sets [10].”
Privacy: Theory meets Practice on the Map (2008, background): “Random sampling [36] has also been analyzed in the differential privacy framework.”
No free lunch in data privacy (2011, background): “Since then, many additional variations of differential privacy have been proposed [9, 4, 29, 2, 23, 20, 28].”
Renyi Differential Privacy Mechanisms for Posterior Sampling (2017, background): “Approximate differential privacy is also related to, but is weaker than, the closely related δ-probabilistic privacy [11] and (1, , δ)-indistinguishability [4].”
The Optimal Mechanism in Differential Privacy : Multidimensional Setting (2013, background): “[14] K. Chaudhuri and N. Mishra, “When random sampling preserves privacy,” in Proceedings of the 26th annual international conference on Advances in Cryptology, ser. CRYPTO’06.”
A rigorous and customizable framework for privacy (2012, background): “The Pufferfish framework differs from differential privacy [12] and its variants [11, 30, 5, 3, 22, 21, 25, 27, 18, 41, 33] in that it can provide precise guarantees about precisely what is being protected and under what conditions the protections hold.”
Sample Complexity Bounds for Differentially Private Learning (2011, background): “There has been a significant amount of work on differential privacy applied to a wide variety of data analysis tasks (Dwork et al., 2006; Chaudhuri and Mishra, 2006; Nissim et al., 2007; Barak et al., 2007; McSherry and Mironov, 2009).”
On sampling, anonymization, and differential privacy or, k-anonymization meets differential privacy (2011, background): “On the negative side, we show that any privacy notion that exploits the adversary’s uncertainty about the data is unlikely to compose, in the sense that publishing the output from two algorithms together may be non-private.”
Near-Optimal Algorithms for Differentially-Private Principal Components (2012, background): “For example, implementing objective perturbation Chaudhuri et al. [2011] uses numerical optimization tools for approximate solutions to convex optimization problems, which have complex termination conditions that are not part of the accompanying theoretical analysis.”
Protecting privacy when mining and sharing user data (2012, background): “Randomly sampling a dataset is known to preserve privacy under a definition closely related to di↵erential privacy [32], but only under the assumption that there are few rare values.”
Pufferfish (2014, background): “P uffer F ish( S , pairs ) satisﬁes the axioms of convexity and transformation invariance.”