Saved in:
Bibliographic Details
Main Authors: Sun, Yu, Xiong, Gaojian, Yao, Xianxun, Ma, Kailang, Cui, Jian
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2401.11748
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909156022157312
author Sun, Yu
Xiong, Gaojian
Yao, Xianxun
Ma, Kailang
Cui, Jian
author_facet Sun, Yu
Xiong, Gaojian
Yao, Xianxun
Ma, Kailang
Cui, Jian
contents Deep gradient inversion attacks expose a serious threat to Federated Learning (FL) by accurately recovering private data from shared gradients. However, the state-of-the-art heavily relies on impractical assumptions to access excessive auxiliary data, which violates the basic data partitioning principle of FL. In this paper, a novel method, Gradient Inversion Attack using Practical Image Prior (GI-PIP), is proposed under a revised threat model. GI-PIP exploits anomaly detection models to capture the underlying distribution from fewer data, while GAN-based methods consume significant more data to synthesize images. The extracted distribution is then leveraged to regulate the attack process as Anomaly Score loss. Experimental results show that GI-PIP achieves a 16.12 dB PSNR recovery using only 3.8% data of ImageNet, while GAN-based methods necessitate over 70%. Moreover, GI-PIP exhibits superior capability on distribution generalization compared to GAN-based methods. Our approach significantly alleviates the auxiliary data requirement on both amount and distribution in gradient inversion attacks, hence posing more substantial threat to real-world FL.
format Preprint
id arxiv_https___arxiv_org_abs_2401_11748
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle GI-PIP: Do We Require Impractical Auxiliary Dataset for Gradient Inversion Attacks?
Sun, Yu
Xiong, Gaojian
Yao, Xianxun
Ma, Kailang
Cui, Jian
Cryptography and Security
Artificial Intelligence
Machine Learning
Deep gradient inversion attacks expose a serious threat to Federated Learning (FL) by accurately recovering private data from shared gradients. However, the state-of-the-art heavily relies on impractical assumptions to access excessive auxiliary data, which violates the basic data partitioning principle of FL. In this paper, a novel method, Gradient Inversion Attack using Practical Image Prior (GI-PIP), is proposed under a revised threat model. GI-PIP exploits anomaly detection models to capture the underlying distribution from fewer data, while GAN-based methods consume significant more data to synthesize images. The extracted distribution is then leveraged to regulate the attack process as Anomaly Score loss. Experimental results show that GI-PIP achieves a 16.12 dB PSNR recovery using only 3.8% data of ImageNet, while GAN-based methods necessitate over 70%. Moreover, GI-PIP exhibits superior capability on distribution generalization compared to GAN-based methods. Our approach significantly alleviates the auxiliary data requirement on both amount and distribution in gradient inversion attacks, hence posing more substantial threat to real-world FL.
title GI-PIP: Do We Require Impractical Auxiliary Dataset for Gradient Inversion Attacks?
topic Cryptography and Security
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2401.11748