Federated Learning with Buffered Asynchronous Aggregation
This work addresses scalability and privacy issues in cross-device federated learning systems, offering a practical solution for large-scale deployments, though it is incremental in improving existing FL methods.
The paper tackles the scalability and privacy trade-offs in federated learning by proposing FedBuff, a buffered asynchronous aggregation method that combines the benefits of synchronous and asynchronous approaches, achieving up to 3.3x efficiency gains over synchronous FL and 2.5x over asynchronous FL while maintaining compatibility with privacy-preserving technologies.
Scalability and privacy are two critical concerns for cross-device federated learning (FL) systems. In this work, we identify that synchronous FL - synchronized aggregation of client updates in FL - cannot scale efficiently beyond a few hundred clients training in parallel. It leads to diminishing returns in model performance and training speed, analogous to large-batch training. On the other hand, asynchronous aggregation of client updates in FL (i.e., asynchronous FL) alleviates the scalability issue. However, aggregating individual client updates is incompatible with Secure Aggregation, which could result in an undesirable level of privacy for the system. To address these concerns, we propose a novel buffered asynchronous aggregation method, FedBuff, that is agnostic to the choice of optimizer, and combines the best properties of synchronous and asynchronous FL. We empirically demonstrate that FedBuff is 3.3x more efficient than synchronous FL and up to 2.5x more efficient than asynchronous FL, while being compatible with privacy-preserving technologies such as Secure Aggregation and differential privacy. We provide theoretical convergence guarantees in a smooth non-convex setting. Finally, we show that under differentially private training, FedBuff can outperform FedAvgM at low privacy settings and achieve the same utility for higher privacy settings.