Membership Inference Attacks Against Time-Series Models

TL;DR

This paper investigates the vulnerability of time-series models, especially in healthcare, to membership inference attacks, introducing new features based on seasonality and trend analysis to improve attack effectiveness.

Contribution

It is the first to adapt and enhance membership inference attacks for time-series models by incorporating seasonality and trend features, revealing privacy risks in medical data applications.

Findings

New features improve attack accuracy on time-series models.

Seasonality and trend analysis enhance membership inference effectiveness.

Medical time-series data privacy risks are significant and quantifiable.

Abstract

Analyzing time-series data that contains personal information, particularly in the medical field, presents serious privacy concerns. Sensitive health data from patients is often used to train machine learning models for diagnostics and ongoing care. Assessing the privacy risk of such models is crucial to making knowledgeable decisions on whether to use a model in production or share it with third parties. Membership Inference Attacks (MIA) are a key method for this kind of evaluation, however time-series prediction models have not been thoroughly studied in this context. We explore existing MIA techniques on time-series models, and introduce new features, focusing on the seasonality and trend components of the data. Seasonality is estimated using a multivariate Fourier transform, and a low-degree polynomial is used to approximate trends. We applied these techniques to various types of time-series models, using datasets from the health domain. Our results demonstrate that these new features enhance the effectiveness of MIAs in identifying membership, improving the understanding of privacy risks in medical data applications.