Privacy-Compatibility For General Utility Metrics

TL;DR

This paper characterizes utility metrics compatible with differential privacy, enabling the design of privacy-preserving mechanisms that maintain meaningful data utility.

Contribution

It provides a complete theoretical characterization of utility metrics that support non-trivial differential privacy guarantees.

Findings

Identifies utility metrics compatible with differential privacy

Provides a theoretical framework for utility metric selection

Enables design of privacy-preserving data mechanisms

Abstract

In this note, we present a complete characterization of the utility metrics that allow for non-trivial differential privacy guarantees.