Smooth Private Forest

Implementation of Smooth Private Forest, a differentially private decision forest designed to minimize the number of queries required and the sensitivity of those queries. Originally published in:

Fletcher, S., & Islam, M. Z. (2017). Differentially private random decision forests using smooth sensitivity. Expert Systems with Applications, 78, 16-31.

BibTeX

@article{fletcher2017differentially,
  title={Differentially private random decision forests using smooth sensitivity},
  author={Fletcher, Sam and Islam, Md Zahidul},
  journal={Expert Systems with Applications},
  volume={78},
  pages={16--31},
  year={2017},
  publisher={Elsevier}
}

Installation

Either download SmoothPrivateForest from the Weka package manager, or download the latest release from the "Releases" section on the sidebar of Github.

Compilation / Development

Set up a project in your IDE of choice, including weka.jar as a compile-time library.

Options

-N <number of trees in forest> Number of trees in forest. (default 10)

-D <number of display trees> Number of trees to display in the output. (default 3)

-E <epsilon> The privacy budget (epsilon) for the exponential mechanism. (default 1.0)

-P Whether or not to display flipped majorities, sensitivity information and true distributions in leaves. (default true)

-C <classname> Specify the full class name of the classifier to compare with Smooth Private Forest.

-S <num> Seed for random number generator. (default 1)