Membership Inference Attacks via Adversarial Examples

TL;DR

This paper investigates membership inference attacks that exploit adversarial examples to assess data privacy risks in machine learning models, and proposes a new defense mechanism supported by empirical results.

Contribution

It introduces a novel method to measure training data leakage using a proxy of total variation and offers a new defense mechanism against such attacks.

Findings

Effective measurement of data leakage via total variation proxy

Demonstrated defense mechanism reduces privacy risks

Empirical evidence confirms attack and defense effectiveness

Abstract

The raise of machine learning and deep learning led to significant improvement in several domains. This change is supported by both the dramatic rise in computation power and the collection of large datasets. Such massive datasets often include personal data which can represent a threat to privacy. Membership inference attacks are a novel direction of research which aims at recovering training data used by a learning algorithm. In this paper, we develop a mean to measure the leakage of training data leveraging a quantity appearing as a proxy of the total variation of a trained model near its training samples. We extend our work by providing a novel defense mechanism. Our contributions are supported by empirical evidence through convincing numerical experiments.