FL Day 2 recap

A visually engaging image depicting federated learning with a focus on client-server and peer-to-peer communication, showcasing data flow and collaboration in technology.

Cross-Silo Federated Learning Quiz

Test your knowledge on cross-silo federated learning and related concepts! This quiz covers various techniques, types of data heterogeneity, and collaborative learning methods.

5 engaging questions
Check your understanding of key concepts
Perfect for students, researchers, and professionals!

5 Questions1 MinutesCreated by LearningTree351

What is typical to cross-silo FL?

Relatively low number of clients

Massive number of clients (>100K)

Relatively few failures

Highly unreliable - 5% or more of clients in a round are expected to fail

Which techniques can be used to reduce communication costs?

Which type of data heterogeneity describes the following example: Images of houses may vary dramatically around the world and in different seasons.

Concept drift - same label different features

Concept shift - same feature different label

Feature distribution skew

Label distribution skew

What is typical to peer-to-peer learning?

There is always a central communication through the server

The clients can communicate with each other

Harder coordination of the process due to the absence of centralised server

What is a split learning?

When clients train models locally and then send them to the server for the aggregation.

When clients send/receive models only to/from each other.

When client train only part of the model, sends “smashed data” to the server which keeps on training the rest layers of the m

FL Day 2 recap

Cross-Silo Federated Learning Quiz

More Surveys