Extracting Spatiotemporal Data from Gradients with Large Language Models

TL;DR

This paper introduces novel gradient inversion attacks and defenses for spatiotemporal federated learning, demonstrating how to reconstruct location data from gradients and proposing adaptive strategies to protect privacy while maintaining utility.

Contribution

The paper develops ST-GIA+ with language model guidance for accurate data reconstruction and proposes an adaptive defense mechanism for enhanced privacy protection.

Findings

ST-GIA+ effectively reconstructs original location data from gradients.

Adaptive defense balances privacy and utility better than existing methods.

Experimental results on real datasets validate the approach's effectiveness.

Abstract

Recent works show that sensitive user data can be reconstructed from gradient updates, breaking the key privacy promise of federated learning. While success was demonstrated primarily on image data, these methods do not directly transfer to other domains, such as spatiotemporal data. To understand privacy risks in spatiotemporal federated learning, we first propose Spatiotemporal Gradient Inversion Attack (ST-GIA), a gradient attack algorithm tailored to spatiotemporal data that successfully reconstructs the original location from gradients. Furthermore, the absence of priors in attacks on spatiotemporal data has hindered the accurate reconstruction of real client data. To address this limitation, we propose ST-GIA+, which utilizes an auxiliary language model to guide the search for potential locations, thereby successfully reconstructing the original data from gradients. In addition, we design an adaptive defense strategy to mitigate gradient inversion attacks in spatiotemporal federated learning. By dynamically adjusting the perturbation levels, we can offer tailored protection for varying rounds of training data, thereby achieving a better trade-off between privacy and utility than current state-of-the-art methods. Through intensive experimental analysis on three real-world datasets, we reveal that the proposed defense strategy can well preserve the utility of spatiotemporal federated learning with effective security protection.