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

2026-03-25

AI Summary

Time series imputation models are widely deployed in healthcare, finance, and other domains, yet their black-box nature poses significant privacy risks, including membership inference and attribute leakage. This work proposes the first two-stage black-box attack framework tailored to such models: it begins with a high-precision membership inference method designed for overfitting-resistant architectures, leveraging reference models to enhance detection performance; it then introduces the first attribute inference attack in the context of time series imputation, uncovering the model’s memorization of sensitive features. Experimental results demonstrate that the proposed approach substantially outperforms baseline methods on the tpr@top25% metric and achieves 90% accuracy in predicting the success of attribute inference—compared to 78% for baselines—thereby effectively establishing a mechanistic link between membership inference and attribute leakage.

Technology Category

• Machine Learning: Privacy
• Computer Vision: Adversarial Attacks & Robustness
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Security and Privacy: Data transparency and provenance
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Deep learning models for time series imputation are now essential in fields such as healthcare, the Internet of Things (IoT), and finance. However, their deployment raises critical privacy concerns. Beyond the well-known issue of unintended memorization, which has been extensively studied in generative models, we demonstrate that time series models are vulnerable to inference attacks in a black-box setting. In this work, we introduce a two-stage attack framework comprising: (1) a novel membership inference attack based on a reference model that improves detection accuracy, even for models robust to overfitting-based attacks, and (2) the first attribute inference attack that predicts sensitive characteristics of the training data for timeseries imputation model. We evaluate these attacks on attention-based and autoencoder architectures in two scenarios: models that are trained from scratch, and fine-tuned models where the adversary has access to the initial weights. Our experimental results demonstrate that the proposed membership attack retrieves a significant portion of the training data with a tpr@top25% score significantly higher than a naive attack baseline. We show that our membership attack also provides a good insight of whether attribute inference will work (with a precision of 90% instead of 78% in the genral case).

Problem

• time series imputation
• privacy
• membership inference
• attribute leakage
• deep learning

Innovation

• membership inference
• attribute leakage
• time series imputation
• privacy attack
• reference-based attack