Horizontally Partitioned Data Publication with Differential Privacy
Algorithm 1
HPDP-DP algorithm.
Input: Data sets , . Private key , public key ., where . Privacy budget and cumulative contribution rate
Output: Synthetic data set
(1)
for to do
(2)
Data owner generates noise matrices and , let and be the symmetric matrix with the upper triangle (including the diagonal) entries are sampled from Gamma , and set .
(3)
Compute:
(4)
Compute:
(5)
for to do
(6)
for to do
(7)
Compute: mod
(8)
end for
(9)
end for
(10)
end for
(11)
return
(12)
Compute the Hadamard product:
(13)
Decrypt :
(14)
Compute:
(15)
Eigenvalue decomposition of matrix , return eigenvalues in descending order , and corresponding eigenvectors
(16)
for to do
(17)
ifthen
(18)
(19)
(20)
end if
(21)
end for
(22)
return
(23)
for to do
(24)
Compute
(25)
Use the model defined in Theorem 1 to generate a synthetic data set
