Table 2.1 A comparison of the key...

Table 2.1

A comparison of the key distinctions between federated learning and fully decentralized learning. Note that as with FL, decentralized learning can be further divided into different use-cases, with distinctions similar to those made in Table 1.1 comparing cross-silo and cross-device FL

	Federated Learning	Fully Decentralized (Peer-to-Peer) Learning
Orchestration	A central orchestration server or service organizes the training, but never sees raw data.	No centralized orchestration.
Wide-area communication	Typically a hub-and-spoke topology, with the hub representing a coordinating service provider (typically without data) and the spokes connecting to clients.	Peer-to-peer topology, with a possibly dynamic connectivity graph.

	Federated Learning	Fully Decentralized (Peer-to-Peer) Learning
Orchestration	A central orchestration server or service organizes the training, but never sees raw data.	No centralized orchestration.
Wide-area communication	Typically a hub-and-spoke topology, with the hub representing a coordinating service provider (typically without data) and the spokes connecting to clients.	Peer-to-peer topology, with a possibly dynamic connectivity graph.

[ViewLarge]

Sharing Unavailable