Membership Inference Attacks via Adversarial Examples
This addresses privacy risks for individuals whose data is used in training datasets, representing an incremental advance in membership inference attack research.
The paper tackles the problem of privacy leakage in machine learning by developing a method to measure training data leakage using a proxy for model variation near training samples, and proposes a defense mechanism, supported by empirical evidence.
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.