Skip to main content

Table 5 The lowest MSE for the existing method and the enhancements which were tested on MS-FIMU dataset

From: Sarve: synthetic data and local differential privacy for private frequency estimation

Method ε = ln(2) ε = ln(3) ε = ln(4) ε = ln(5) ε = ln(6) ε = ln(7)
RS + FD[ADP] 0.000108 5.59E05 4.64E05 4.40E05 3.37E−05 2.97E−05
RAPPOR 0.000207 8.86E−05 8.01E−05 5.29E−05 5.78E−05 4.75E−05
Hadamard response 0.004249 0.001295 0.000635 0.00037 0.000263 0.000187
Sarve 0.000105 6.73E−05 4.78E−05 4.41E−05 3.20E05 2.72E05
Method ε = 2 ε = 3 ε = 4 ε = 5 ε = 6 ε = 7
RS + FD[ADP] 2.02E05 1.12E−05 5.87E06 3.90E06 4.35E06 3.10E06
RAPPOR 4.87E−05 2.28E−05 1.23E−05 8.58E−06 5.73E−06 4.91E−06
Hadamard response 0.000164 3.79E−05 1.88E−05 1.13E−05 1.72E−05 1.78E−05
Sarve 2.50E−05 9.78E06 6.81E−06 4.64E−06 4.66E−06 3.21E−06
  1. The values in bold indicate privacy conditions when Sarve performed better than adaptive RS+FD and resulted in lower MSE between real and post-privatization estimated frequencies