Embedding-Space Data Augmentation to Prevent Membership Inference Attacks in Clinical Time Series Forecasting

Marius Fracarolli, Michael Staniek, Stefan Riezler
11/7/2025
cs.LG

No official code found. Best implementation: MariusFracarolli/ML4H_2025_Data-Augmentation-to-Prevent-MIA-in-TSF (1 star, last commit 2025-10-31).

Abstract

Balancing strong privacy guarantees with high predictive performance is critical for time series forecasting (TSF) tasks involving Electronic Health Records (EHR). In this study, we explore how data augmentation can mitigate Membership Inference Attacks (MIA) on TSF models. We show that retraining with synthetic data can substantially reduce the effectiveness of loss-based MIAs by reducing the attacker's true-positive to false-positive ratio. The key challenge is generating synthetic samples that closely resemble the original training data to confuse the attacker, while also introducing enough novelty to enhance the model's ability to generalize to unseen data. We examine multiple augmentation strategies - Zeroth-Order Optimization (ZOO), a variant of ZOO constrained by Principal Component Analysis (ZOO-PCA), and MixUp - to strengthen model resilience without sacrificing accuracy. Our experimental results show that ZOO-PCA yields the best reductions in TPR/FPR ratio for MIA attacks without sacrificing performance on test data.

View on arXivView PDF

Code Implementations(1)

MariusFracarolli/ML4H_2025_Data-Augmentation-to-Prevent-MIA-in-TSFCommunity65%

Embedding-Space Data Augmentation to Prevent Membership Inference Attacks in Clinical Time Series Forecasting, published at ML4H 2025

10PythonOct 31, 202511 months ago

Similar Papers

Privacy Risks in Time Series Forecasting: User- and Record-Level Membership Inference

87% match2025

Membership Inference Attacks Against Time-Series Models

87% match2024

A new membership inference attack that spots memorization in generative and predictive models: Loss-Based with Reference Model algorithm (LBRM)

83% match2025

Uncovering Memorization in Timeseries Imputation models: LBRM Membership Inference and its link to attribute Leakage

82% match2026

In-Context Probing for Membership Inference in Fine-Tuned Language Models

82% match1 repo122025

Overconfidence is a Dangerous Thing: Mitigating Membership Inference Attacks by Enforcing Less Confident Prediction

81% match1 repo132023

Cite this paper

@article{fracarolli2025embeddingspace,
  title  = {Embedding-Space Data Augmentation to Prevent Membership Inference Attacks in Clinical Time Series Forecasting},
  author = {Marius Fracarolli and Michael Staniek and Stefan Riezler},
  year   = {2025},
  eprint = {2511.05289},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2511.05289v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2511.05289.svg)](https://codeofpaper.com/paper/2511.05289)

Discussion