Parameter Exploration Tool

This tool implements a count query perturbation mechanism for the preservation of privacy. See here for background and availability.

Probablity and Utility Functions

U(r,c) = b|r - c|^a

Parameters

Double-click values to edit or select a preset, then click "Recompute".

c	ε	α⁺	β⁺	α^-	β^-	r_min	r_max	n

Presets:

Computed values

A random deviate r is chosen with probability P(r|c). Let Δ be the sensitivity of U, i.e. the max change in U for any r between r_min and r_max resulting from a unit change in c for c between 0 and n. Then we have that releasing r affords ε = 2ηΔ differential privacy. For a chosen ε, we compute the needed η. Furthermore, we compute the discrete distribution P(r|c) and its mean and variance. Finally, a few random deviates are chosen using P(r|c). Click the "Recompute" button to generate a new random deviates and refresh the plots.

η	Variance	Mean	Random Deviates

Utility Shape

Probability Mass

Background and Availability

Today’s clinical research institutions provide tools for researchers to query their data warehouses for counts of patients. This tool implements a perturbation mechanism in which users are given options with respect to scale and direction of the perturbation. It is intended to help system designers and users to visually and numerically investigate the interaction between system parameters, including required privacy level, and user preference settings. The source code is distributed under of the GNU General Public License version 2 or higher, and can be found here. A full description of the mechanism can be found in:

Protecting count queries in study design.
Vinterbo SA, Sarwate AD, Boxwala AA.
J Am Med Inform Assoc. 2012 Sep 1;19(5):750-7. Epub 2012 Apr 17.

This implementation was in part supported by NIH NLM grant 7R01LM007273-07 and NIH Roadmap for Medical Research grant U54 HL108460.