Differentially Private Federated Learning via Inexact ADMM
This work addresses privacy concerns in federated learning for applications like image classification, though it is incremental as it builds on existing DP and ADMM techniques.
The paper tackles the trade-off between data privacy and learning performance in federated learning by developing a differentially private inexact ADMM algorithm, which reduces testing error by up to 22% compared to existing methods while maintaining the same privacy level and converging faster.
Differential privacy (DP) techniques can be applied to the federated learning model to protect data privacy against inference attacks to communication among the learning agents. The DP techniques, however, hinder achieving a greater learning performance while ensuring strong data privacy. In this paper we develop a DP inexact alternating direction method of multipliers algorithm that solves a sequence of subproblems with the objective perturbation by random noises generated from a Laplace distribution. We show that our algorithm provides $\barε$-DP for every iteration, where $\barε$ is a privacy parameter controlled by a user. Using MNIST and FEMNIST datasets for the image classification, we demonstrate that our algorithm reduces the testing error by at most $22\%$ compared with the existing DP algorithm, while achieving the same level of data privacy. The numerical experiment also shows that our algorithm converges faster than the existing algorithm.