FetchSGD: Communication-Efficient Federated Learning with Sketching
This addresses communication efficiency and convergence problems in federated learning for distributed machine learning systems, representing a novel method rather than an incremental improvement.
The paper tackles the communication bottleneck and convergence issues in federated learning due to sparse client participation by introducing FetchSGD, which compresses model updates using a Count Sketch and moves momentum and error accumulation to the central aggregator, achieving high compression rates and good convergence as demonstrated by training residual networks and a transformer model.
Existing approaches to federated learning suffer from a communication bottleneck as well as convergence issues due to sparse client participation. In this paper we introduce a novel algorithm, called FetchSGD, to overcome these challenges. FetchSGD compresses model updates using a Count Sketch, and then takes advantage of the mergeability of sketches to combine model updates from many workers. A key insight in the design of FetchSGD is that, because the Count Sketch is linear, momentum and error accumulation can both be carried out within the sketch. This allows the algorithm to move momentum and error accumulation from clients to the central aggregator, overcoming the challenges of sparse client participation while still achieving high compression rates and good convergence. We prove that FetchSGD has favorable convergence guarantees, and we demonstrate its empirical effectiveness by training two residual networks and a transformer model.