Differentially Private High-Dimensional Data Publication via Sampling-Based Inference

By Rui Chen, Qian Xiao, Yu Zhang, Jianliang Xu. 2015

206 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Science Applications

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 Computer Science Applications

Citations in the corpus

Detecting Violations of Differential Privacy (2018, methodology): “0) [1, 1, 1, 1, 1] [2, 1, 1, 1, 1] iSVT 1 [38] t (ω ) = 0 [1, 1, 1, 1, 1, 1, 1, 1, 1, 1] [0, 0, 0, 0, 0, 2, 2, 2, 2, 2] iSVT 2 [11] t (ω ) = 9 [1, 1, 1, 1, 1, 0, 0, 0, 0, 0] [0, 0, 0, 0, 0, 1, 1, 1, 1, 1] iSVT 3 [27] t (ω ) = 0 [1, 1, 1, 1, 1, 0, 0, 0, 0, 0] [0, 0, 0, 0, 0, 1, 1, 1, 1, 1] iSVT 4 [36”
Enhancing Transparency and Control When Drawing Data-Driven Inferences About Individuals (2016, background): “Chen et al.(30) discuss differentially private high-dimensional data publication.”
Privacy-Preserving Collaborative Deep Learning with Irregular Participants. (2018, background): “Differential privacy has become a de facto standard privacy model for statistics analysis with provable privacy guarantee, and has been widely used in data publishing [32], [33] and data analysis [34], [35].”
$\textsf{LoPub}$ : High-Dimensional Crowdsourced Data Publication With Local Differential Privacy (2016, methodology): “Comparison LoPub (Our method) RAPPOR [12] EM [14] JTree [5] Local privacy Y Y Y N High Dimension Y N N Y Communication O( ∑ j |Ωj|) O( ∏ j |Ωj |) O( ∑ j |Ωj|) -”
Differentially Private Data Publishing and Analysis: A Survey (2017, methodology): “Even though [50] and [51] have undertaken some initial work, they both consume too much of the privacy budget when building the distributionmodel, making the results less accurate than that of lower-dimensional datasets.”
Machine Learning for Synthetic Data Generation: a Review (2023, unknown): “Regulated data such as clinical and genomics data in raw may not be shared and one solution is to share synthesized data instead.”
PrivBayes: private data release via bayesian networks (2014, background): “This invalidates the solution proposed in [9], and we have not seen any easy way to address this issue.”
Answering Multi-Dimensional Analytical Queries under Local Differential Privacy (2019, result): “As shown in Figure 3, [ 2 , 7 ] can be partitioned into 4 intervals in I D (blue ones), and [ 3 , 8 ] partitioned into 2 in I D (green ones).”
CALM: Consistent Adaptive Local Marginal for Marginal Release under Local Differential Privacy (2018, background): “Note that the problem has also been investigated in the centralized setting [10, 29, 41], but the techniques cannot be directly used in this setting, because there is no centralized aggregator that has the overall view of all users’ data, and the noise in the local setting is significantly larger…”
Locally Differentially Private Data Collection and Analysis (2019, background): “Since the introduction of differential privacy (DP), a large number of DP-based mechanisms [27, 25, 57] have been proposed and applied to numerous scenarios, such as data statistics [54, 61, 11], learning models [1, 40, 59, 38], and systems [29, 6].”
Differential Privacy Techniques for Cyber Physical Systems: A Survey (2018, background): “Similarly, in the domain of big data, maintaining privacy protection along with issue of dimensionality because of large data volume and computation overhead is a big challenge for researchers [335], [336].”
How to DP-fy ML: A Practical Guide to Machine Learning with Differential Privacy (2023, methodology): “JunctionTree method (Chen et al., 2015) subsequently improved upon PrivBayes by learning DP-protected pairwise correlations of the attributes and applying junction tree algorithm to infer joint data distribution via noisy marginals.”
On the Privacy Properties of Variants on the Sparse Vector Technique (2015, methodology): “Chen et al [2] instantiate GPTT with ǫ 1 = ǫ 2 = ǫ 2 , for generating synthetic data.”