DP.Sim: Releasing Differential Private RKHS smoothing mean of a...

In sxz155/PFDA:

Description

This function create a DP RKHS smoothing mean from an existing data set with known eigenvalues and eigenvectors

Usage

DP.Sim(alpha = 1, beta = 0.1, kernel = "Gau", phi = 0.01, ro = 0.2,
  n = 100, N = 25, tau = 0.4, case = 2, pow = 4, mu = rep(0, n))

Arguments

alpha, beta	Privacy parameters, real numbers
phi	real number, penalty parameter
ro	range parameter in kernel, real number
n	real number, number of grid points
N	real number, number of observations
tau	range of the uniform distribution in KL expansion
pow	smoothing parameter, e.val.x_i=i^-pow
mu	real vector n*1, initial mean vector
e.val.x	real vector n*1, eigenvalues
e.vec.x	real valued matrix n*N, eigenvectors
e.val.z	real valued matrix n*N, eigenvectors of noise

Value

f.tilda: DP RKHS smoothing mean, n*1 real valued vector

delta: the coefficient of the noise, real number

f: RKHS smoothing mean, n*1 real valued vector

X: original data generated by the selected noise covariance operator C by KL epansion when e.val.x_i=i^-pow and e.vec.x=e.vec.z ...

sxz155/PFDA documentation built on May 30, 2019, 10:40 p.m.