pooledCovDataAccess {DPpack} | R Documentation |
Differentially Private Pooled Covariance Data Access Function
Description
This function performs the data access step in the computation of a
differentially private pooled covariance. The true values are computed using
the theoretical formula and cov
, while the sensitivities
are calculated based on bounded and unbounded differential privacy
(Kifer and Machanavajjhala 2011) according to the theoretical values
(Liu 2019).
Usage
pooledCovDataAccess(
samples,
lower.bound1,
upper.bound1,
lower.bound2,
upper.bound2,
approx.n.max
)
Arguments
samples |
List of two-column matrices from which to compute the pooled covariance. |
lower.bound1 , lower.bound2 |
Real numbers giving the lower bounds of the first and second columns of samples, respectively. |
upper.bound1 , upper.bound2 |
Real numbers giving the upper bounds of the first and second columns of samples, respectively. |
approx.n.max |
Logical indicating whether to approximate n.max, which is defined to be the length of the largest input vector. Approximation is best if n.max is very large. |
Value
List of the true pooled covariance and the sensitivities calculated based on bounded and unbounded differential privacy.
References
Liu F (2019). “Statistical Properties of Sanitized Results from Differentially Private Laplace Mechanism with Univariate Bounding Constraints.” Transactions on Data Privacy, 12(3), 169-195. http://www.tdp.cat/issues16/tdp.a316a18.pdf.
Kifer D, Machanavajjhala A (2011). “No Free Lunch in Data Privacy.” In Proceedings of the 2011 ACM SIGMOD International Conference on Management of Data, SIGMOD '11, 193–204. ISBN 9781450306614, doi:10.1145/1989323.1989345.
Examples
x1 <- matrix(c(1,4,-2,8,-6,-3),ncol=2)
x2 <- matrix(c(1,2,-5,7),ncol=2)
pooledCovDataAccess(list(x1,x2),-10,10,-10,10,FALSE)