Increasing the accuracy of federated learning on non-iid scenarios using client clustering

HIGHLIGHTS

• What: The authors propose clustering system to mitigate the convergence obstacles of federated learning in nonIndependent and Identically Distributed (IID) scenarios. The authors propose a system that achieves high accuracy results under non-IID data distributions. The authors implement a proof-of-concept of the proposed system. The authors propose a clustering system, dividing FL clients into groups with an unsupervised clustering algorithm.
• Who: Lucas Airam Castro de Souza and collaborators from the Universidade Federal do Rio Janeiro (UFRJ), Brazil, have published the article: Increasing the Accuracy of Federated Learning on Non-IID Scenarios using . . .
  

   

  

  If you want to have access to all the content you need to log in!

  Thanks :)

  Username or Email

  Password

  Remember me

  Lost your password?

  If you don't have an account, you can create one here.