Adaptive Federated Optimization
This work addresses convergence issues in federated learning for distributed machine learning applications, but it is incremental as it adapts existing non-federated methods to the federated setting.
The paper tackled the problem of difficult tuning and unfavorable convergence in federated learning by proposing federated versions of adaptive optimizers like Adagrad, Adam, and Yogi, and showed that these methods significantly improve performance through extensive experiments.
Federated learning is a distributed machine learning paradigm in which a large number of clients coordinate with a central server to learn a model without sharing their own training data. Standard federated optimization methods such as Federated Averaging (FedAvg) are often difficult to tune and exhibit unfavorable convergence behavior. In non-federated settings, adaptive optimization methods have had notable success in combating such issues. In this work, we propose federated versions of adaptive optimizers, including Adagrad, Adam, and Yogi, and analyze their convergence in the presence of heterogeneous data for general non-convex settings. Our results highlight the interplay between client heterogeneity and communication efficiency. We also perform extensive experiments on these methods and show that the use of adaptive optimizers can significantly improve the performance of federated learning.