Timezone: »

 
Poster
Throughput-Optimal Topology Design for Cross-Silo Federated Learning
Othmane Marfoq · CHUAN XU · Giovanni Neglia · Richard Vidal

Tue Dec 08 09:00 AM -- 11:00 AM (PST) @ Poster Session 1 #328

Federated learning usually employs a client-server architecture where an orchestrator iteratively aggregates model updates from remote clients and pushes them back a refined model. This approach may be inefficient in cross-silo settings, as close-by data silos with high-speed access links may exchange information faster than with the orchestrator, and the orchestrator may become a communication bottleneck. In this paper we define the problem of topology design for cross-silo federated learning using the theory of max-plus linear systems to compute the system throughput---number of communication rounds per time unit. We also propose practical algorithms that, under the knowledge of measurable network characteristics, find a topology with the largest throughput or with provable throughput guarantees. In realistic Internet networks with 10~Gbps access links for silos, our algorithms speed up training by a factor 9 and 1.5 in comparison to the master-slave architecture and to state-of-the-art MATCHA, respectively. Speedups are even larger with slower access links.

Author Information

Othmane Marfoq (Inria / Accenture)
CHUAN XU (Inria Sophia Antipolis)
Giovanni Neglia (Inria)
Richard Vidal (Accenture)

More from the Same Authors