Enhanced Privacy Leakage from Noise-Perturbed Gradients via Gradient-Guided Conditional Diffusion Models

TL;DR

This paper introduces a novel gradient-guided diffusion model attack that effectively reconstructs private images from noisy gradients in federated learning, surpassing existing methods and analyzing factors affecting attack success.

Contribution

The paper presents a new diffusion-based attack method that improves privacy leakage from noisy gradients without prior data distribution knowledge.

Findings

Outperforms existing gradient inversion attacks under noise defenses.

Theoretical bounds on reconstruction error and convergence are established.

Experimental results confirm the effectiveness of the proposed method.

Abstract

Federated learning synchronizes models through gradient transmission and aggregation. However, these gradients pose significant privacy risks, as sensitive training data is embedded within them. Existing gradient inversion attacks suffer from significantly degraded reconstruction performance when gradients are perturbed by noise-a common defense mechanism. In this paper, we introduce gradient-guided conditional diffusion models for reconstructing private images from leaked gradients, without prior knowledge of the target data distribution. Our approach leverages the inherent denoising capability of diffusion models to circumvent the partial protection offered by noise perturbation, thereby improving attack performance under such defenses. We further provide a theoretical analysis of the reconstruction error bounds and the convergence properties of the attack loss, characterizing the impact of key factors-such as noise magnitude and attacked model architecture-on reconstruction quality. Extensive experiments demonstrate our attack's superior reconstruction performance with Gaussian noise-perturbed gradients, and confirm our theoretical findings.