Asynchronous Federated Optimization
Cong Xie, Sanmi Koyejo, Indranil Gupta · arXiv (Cornell University) · 2019
Federated learning enables training on a massive number of edge devices. To improve flexibility and scalability, we propose a new asynchronous federated optimization algorithm. We prove that the proposed approach has near-linear convergence to a global optimum, for both strongly convex and a restricted family of non-convex problems. Empirical results show that the proposed algorithm converges quickly and tolerates staleness in various applications.
Cited by 410 works. doi:10.48550/arxiv.1903.03934 Open-access PDF
Authors
References
- Learning Multiple Layers of Features from Tiny Images
- MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
- Communication-Efficient Learning of Deep Networks from Decentralized Data
- Federated Learning: Strategies for Improving Communication Efficiency
- Practical Secure Aggregation for Privacy-Preserving Machine Learning
- MXNet: A Flexible and Efficient Machine Learning Library for Heterogeneous Distributed Systems
- Slow Learners are Fast
- Federated Learning of Deep Networks using Model Averaging
- Slow Learners are Fast
- Scaling distributed machine learning with the parameter server