Recovering Private Text in Federated Learning of Language Models

Gupta, Samyak; Huang, Yangsibo; Zhong, Zexuan; Gao, Tianyu; Li, Kai; Chen, Danqi

doi:10.48550/arxiv.2205.08514

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Consistent with related work (Geiping et al 2020;Enthoven and Al-Ars 2020;Yin et al 2021;Kaissis et al 2021;Jin et al 2021;Scheliga, Mäder, and Seeland 2022b,a;Zhang et al 2022;Gupta et al 2022), we assume a honest-butcurious server threat model. In this scenario the attacker has insight into the training process, i.e.…”

Section: Gradient Inversion Attacksmentioning

confidence: 90%

Dropout Is NOT All You Need to Prevent Gradient Leakage

Scheliga

Maeder

Seeland

2023

AAAI

View full text Add to dashboard Cite

Gradient inversion attacks on federated learning systems reconstruct client training data from exchanged gradient information. To defend against such attacks, a variety of defense mechanisms were proposed. However, they usually lead to an unacceptable trade-off between privacy and model utility. Recent observations suggest that dropout could mitigate gradient leakage and improve model utility if added to neural networks. Unfortunately, this phenomenon has not been systematically researched yet. In this work, we thoroughly analyze the effect of dropout on iterative gradient inversion attacks. We find that state of the art attacks are not able to reconstruct the client data due to the stochasticity induced by dropout during model training. Nonetheless, we argue that dropout does not offer reliable protection if the dropout induced stochasticity is adequately modeled during attack optimization. Consequently, we propose a novel Dropout Inversion Attack (DIA) that jointly optimizes for client data and dropout masks to approximate the stochastic client model. We conduct an extensive systematic evaluation of our attack on four seminal model architectures and three image classification datasets of increasing complexity. We find that our proposed attack bypasses the protection seemingly induced by dropout and reconstructs client data with high fidelity. Our work demonstrates that privacy inducing changes to model architectures alone cannot be assumed to reliably protect from gradient leakage and therefore should be combined with complementary defense mechanisms.

show abstract