Use the Spear as a Shield: A Novel Adversarial Example based Privacy-Preserving Technique against Membership Inference Attacks

TL;DR

This paper introduces AEPPT, a novel adversarial perturbation technique that protects machine learning models from membership inference attacks without affecting model accuracy, by misleading the attacker's inference model.

Contribution

The paper presents a general, output-only modification method that effectively defends against membership inference attacks, including adaptive ones, outperforming existing defenses.

Findings

Reduces membership inference accuracy and precision to near 50%.

Effective against adaptive attacks with knowledge of the defense.

Does not compromise the target model's performance.

Abstract

Recently, the membership inference attack poses a serious threat to the privacy of confidential training data of machine learning models. This paper proposes a novel adversarial example based privacy-preserving technique (AEPPT), which adds the crafted adversarial perturbations to the prediction of the target model to mislead the adversary's membership inference model. The added adversarial perturbations do not affect the accuracy of target model, but can prevent the adversary from inferring whether a specific data is in the training set of the target model. Since AEPPT only modifies the original output of the target model, the proposed method is general and does not require modifying or retraining the target model. Experimental results show that the proposed method can reduce the inference accuracy and precision of the membership inference model to 50%, which is close to a random guess. Further, for those adaptive attacks where the adversary knows the defense mechanism, the proposed AEPPT is also demonstrated to be effective. Compared with the state-of-the-art defense methods, the proposed defense can significantly degrade the accuracy and precision of membership inference attacks to 50% (i.e., the same as a random guess) while the performance and utility of the target model will not be affected.