Differentially Private Federated Learning on Heterogeneous Data
arXiv:2111.09278
Abstract
Federated Learning (FL) is a paradigm for large-scale distributed learning which faces two key challenges: (i) efficient training from highly heterogeneous user data, and (ii) protecting the privacy of participating users. In this work, we propose a novel FL approach (DP-SCAFFOLD) to tackle these two challenges together by incorporating Differential Privacy (DP) constraints into the popular SCAFFOLD algorithm. We focus on the challenging setting where users communicate with a "honest-but-curious" server without any trusted intermediary, which requires to ensure privacy not only towards a third-party with access to the final model but also towards the server who observes all user communications. Using advanced results from DP theory, we establish the convergence of our algorithm for convex and non-convex objectives. Our analysis clearly highlights the privacy-utility trade-off under data heterogeneity, and demonstrates the superiority of DP-SCAFFOLD over the state-of-the-art algorithm DP-FedAvg when the number of local updates and the level of heterogeneity grow. Our numerical results confirm our analysis and show that DP-SCAFFOLD provides significant gains in practice.
References in corpus (12)
- Adam: A Method for Stochastic Optimization
- Deep Learning with Differential Privacy
- Communication-Efficient Learning of Deep Networks from Decentralized Data
- Rényi Divergence and Kullback-Leibler Divergence
- Learning Differentially Private Recurrent Language Models
- Inverting Gradients -- How easy is it to break privacy in federated learning?
- A Field Guide to Federated Optimization
- Federated Learning Based on Dynamic Regularization
- Mime: Mimicking Centralized Stochastic Algorithms in Federated Learning
- Differentially Private Learning with Adaptive Clipping
- Local Privacy, Data Processing Inequalities, and Statistical Minimax Rates
- Shuffled Model of Federated Learning: Privacy, Communication and Accuracy Trade-offs