Bayesian Framework for Gradient Leakage
This work addresses privacy vulnerabilities in federated learning for users and developers, but it is incremental as it formalizes and analyzes existing attacks rather than introducing new defenses.
The authors tackled the problem of gradient leakage in federated learning, which compromises data privacy, by proposing a Bayesian framework to analyze the Bayes optimal adversary, and found that existing defenses are ineffective against stronger attacks, especially early in training.
Federated learning is an established method for training machine learning models without sharing training data. However, recent work has shown that it cannot guarantee data privacy as shared gradients can still leak sensitive information. To formalize the problem of gradient leakage, we propose a theoretical framework that enables, for the first time, analysis of the Bayes optimal adversary phrased as an optimization problem. We demonstrate that existing leakage attacks can be seen as approximations of this optimal adversary with different assumptions on the probability distributions of the input data and gradients. Our experiments confirm the effectiveness of the Bayes optimal adversary when it has knowledge of the underlying distribution. Further, our experimental evaluation shows that several existing heuristic defenses are not effective against stronger attacks, especially early in the training process. Thus, our findings indicate that the construction of more effective defenses and their evaluation remains an open problem.