Communication-Efficient Distributed Learning with Differential Privacy
This addresses the problem of balancing communication costs and privacy in distributed learning for applications like federated learning, though it appears incremental as it combines existing techniques like local training and differential privacy.
The paper tackles nonconvex learning over networks by developing an algorithm that is communication-efficient and ensures data privacy through local training with gradient perturbation, proving convergence to a stationary point and showing superior performance on a classification task under the same privacy budget compared to state-of-the-art methods.
We address nonconvex learning problems over undirected networks. In particular, we focus on the challenge of designing an algorithm that is both communication-efficient and that guarantees the privacy of the agents' data. The first goal is achieved through a local training approach, which reduces communication frequency. The second goal is achieved by perturbing gradients during local training, specifically through gradient clipping and additive noise. We prove that the resulting algorithm converges to a stationary point of the problem within a bounded distance. Additionally, we provide theoretical privacy guarantees within a differential privacy framework that ensure agents' training data cannot be inferred from the trained model shared over the network. We show the algorithm's superior performance on a classification task under the same privacy budget, compared with state-of-the-art methods.