Secure Regression on Distributed Databases

By Alan F. Karr, Xiaodong Lin, Ashish P. Sanil, Jerome P. Reiter. 2005

146 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Networks and Communications, Statistics and Probability

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 Networks and Communications

In Statistics and Probability

Citations in the corpus

Prio: Private, Robust, and Scalable Computation of Aggregate Statistics (2017, methodology): “[82] that a system for computing private sums can also privately train linear models.”
Privacy-Preserving Ridge Regression on Hundreds of Millions of Records (2013, background): “Some approaches use protocols based on secret sharing [8], [45]– [47] while others use additive homomorphic encryption (for horizontally partitioned data) [48].”
Privacy preserving regression modelling via distributed computation (2004, background): “…versions of secure summation that involve each agency partitioning their constribution to the sum and all agencies conducting multiple rounds of the summation process to obtain the required sum [4, 9] (the order of the agencies could also be permuted to add an extra layer of security) 2 .”
Privacy-preserving SVM classification (2008, unknown): “Specifically , secure methods have been proposed for association rule mining [18, 36], clustering[37, 38, 39], classification [40, 41, 42, 43], outlier detection [44] and regression[ 45 , 46].”
Privacy-Preserving SVM Classification on Vertically Partitioned Data (2006, background): “In terms of data mining problems, work addressed includes association rule mining [15], clustering [16, 17], classification [18], and regression [ 19 , 20].”
Privacy-Preserving Analysis of Vertically Partitioned Data Using Secure Matrix Products (2009, background): “Key words: Distributed databases; secure matrix product; vertically partitioned data; regression; data confidentiality.”
Fast, Privacy Preserving Linear Regression over Distributed Datasets based on Pre-Distributed Data (2015, background): “Most of them clearly do not even aim at obtaining the level of privacy usually required by modern cryptographic protocols (such as Karr et al. [21] and Du et al.[14], see also [29, 22]).”
Distributed learning: Developing a predictive model based on data from multiple hospitals without data leaving the hospital - A real life proof of concept. (2016, background): “Some of these techniques focus on horizontal distribution of data, meaning that each center has the same variables but different subjects [11,12].”
Secure multiple linear regression based on homomorphic encryption (2011, methodology): “This is the setting explored in for instance, Karr et al. (2009). Although the protocol we proposed above computes the regression coefficients directly, note that it is trivially modified to reveal the statistics {X X, X y} instead, so that the parties themselves may carry out the remaining analysis”
Secure, Privacy-Preserving Analysis of Distributed Databases (2007, methodology): “These include the coefﬁcient of determination R 2 , the least squares estimate S 2 of the error variance σ 2 and the “hat” matrix H = X ( X T X ) 1 X T , which can be used to identify outliers (Karr et al., 2005b,c).”
Privacy-Preserving Record Linkage (2010, background): “For example, numerous papers already tacitly assume that the files that are input to their procedures are a-priori matched, in the sense that the correspondence between the units is known [24, 23, 22, 12, 13].”
Regression on Distributed Databases via Secure Multi-Party Computation (2004, methodology): “Our approach has the additional advantage of being resistant to source identification via attribute values, as discussed in Karr et al. (2004): only data summaries, not data values, are shared.”
VERTIcal Grid lOgistic regression (VERTIGO) (2016, background): “: (2) Figure 1: Horizontally and vertically partitioned data.”