Selects an output with probability proportional to
exp(epsilon u(D, r) / (2 Delta u)) (McSherry & Talwar 2007; Dwork
& Roth 2014, Definition 3.4). Entry point onto this package's
native morie_dp_exponential_mechanism.
References
McSherry, F. and Talwar, K. (2007). Mechanism design via differential privacy. FOCS 2007, 94-103.
Examples
set.seed(26)
morie_expmc(candidates = c("a", "b", "c"), utility = c(1, 5, 2), sensitivity = 1, epsilon = 1)
#> $selected
#> [1] "b"
#>
#> $index
#> [1] 1
#>
#> $probabilities
#> [1] 0.09962365 0.73612472 0.16425163
#>
#> $epsilon
#> [1] 1
#>
#> $sensitivity
#> [1] 1
#>
#> $mechanism
#> [1] "exponential"
#>
#> $method
#> [1] "dp_exponential_mechanism"
#>