Gradient Inversion Transcript: Leveraging Robust Generative Priors to Reconstruct Training Data from Gradient Leakage

TL;DR

This paper introduces GIT, a generative method that effectively reconstructs training data from leaked gradients, outperforming existing techniques and demonstrating robustness across diverse scenarios.

Contribution

GIT is a novel generative approach tailored for gradient inversion, improving reconstruction quality and efficiency compared to prior methods.

Findings

Outperforms existing gradient inversion methods on multiple datasets.

Enhances reconstruction quality by integrating GIT as a prior.

Demonstrates robustness under gradient inaccuracies and data shifts.

Abstract

We propose Gradient Inversion Transcript (GIT), a novel generative approach for reconstructing training data from leaked gradients. GIT employs a generative attack model, whose architecture is tailored to align with the structure of the leaked model based on theoretical analysis. Once trained offline, GIT can be deployed efficiently and only relies on the leaked gradients to reconstruct the input data, rendering it applicable under various distributed learning environments. When used as a prior for other iterative optimization-based methods, GIT not only accelerates convergence but also enhances the overall reconstruction quality. GIT consistently outperforms existing methods across multiple datasets and demonstrates strong robustness under challenging conditions, including inaccurate gradients, data distribution shifts and discrepancies in model parameters.