Understanding Compressive Adversarial Privacy

TL;DR

This paper introduces a compressive adversarial privacy framework that balances data utility and privacy, using convex optimization and neural networks to model data sharing and attack strategies.

Contribution

It presents a novel framework combining linear and nonlinear compression models for privacy, with convex optimization and neural network-based attacks.

Findings

The framework effectively preserves sensitive information in empirical tests.

Nonlinear compression models enhance privacy-utility trade-offs.

Convex optimization characterizes optimal data release mechanisms.

Abstract

Designing a data sharing mechanism without sacrificing too much privacy can be considered as a game between data holders and malicious attackers. This paper describes a compressive adversarial privacy framework that captures the trade-off between the data privacy and utility. We characterize the optimal data releasing mechanism through convex optimization when assuming that both the data holder and attacker can only modify the data using linear transformations. We then build a more realistic data releasing mechanism that can rely on a nonlinear compression model while the attacker uses a neural network. We demonstrate in a series of empirical applications that this framework, consisting of compressive adversarial privacy, can preserve sensitive information.