A Novel Approach to Differential Privacy with Alpha Divergence

TL;DR

This paper introduces alpha differential privacy (ADP), a new privacy framework based on alpha divergence, offering more adaptable privacy guarantees compared to traditional methods, especially in scenarios demanding strict privacy.

Contribution

The paper presents the theoretical foundation of ADP and demonstrates its improved privacy performance over existing frameworks through empirical evaluations.

Findings

ADP provides stronger privacy guarantees in moderate iteration scenarios.

Empirical results show ADP outperforms traditional differential privacy in strict privacy settings.

ADP offers a flexible privacy assessment suitable for modern data analysis environments.

Abstract

As data-driven technologies advance swiftly, maintaining strong privacy measures becomes progressively difficult. Conventional $(\epsilon, \delta)$-differential privacy, while prevalent, exhibits limited adaptability for many applications. To mitigate these constraints, we present alpha differential privacy (ADP), an innovative privacy framework grounded in alpha divergence, which provides a more flexible assessment of privacy consumption. This study delineates the theoretical underpinnings of ADP and contrasts its performance with competing privacy frameworks across many scenarios. Empirical assessments demonstrate that ADP offers enhanced privacy guarantees in small to moderate iteration contexts, particularly where severe privacy requirements are necessary. The suggested method markedly improves privacy-preserving methods, providing a flexible solution for contemporary data analysis issues in a data-centric environment.