Abstract:

The differential privacy model until today was extended and implemented by many researchers. One of the recent extensions proposed two metrics that are designed to allow for better calibration of the security and accuracy based on the risk of the statistical query. The quality of this extension shows that the method as defined can be considered as a better candidate than the original version in certain conditions. This paper further improves the extended differential privacy model with selected aspects of the implementation of the mechanism that may contribute to improving the performance and security of the solution at a system level.