English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Statistical heterogeneity of data present at client devices in a federatedlearning (FL) system renders the training of a global model in such systemsdifficult. Particularly challenging are the settings where due to resourceconstraints only a small fraction of clients can participate in any given roundof FL. Recent approaches to training a global model in FL systems with non-IIDdata have focused on developing client selection methods that aim to sampleclients with more informative updates of the model. However, existing clientselection techniques either introduce significant computation overhead orperform well only in the scenarios where clients have data with similarheterogeneity profiles. In this paper, we propose HiCS-FL (Federated Learningvia Hierarchical Clustered Sampling), a novel client selection method in whichthe server estimates statistical heterogeneity of a client's data using theclient's update of the network's output layer and relies on this information tocluster and sample the clients. We analyze the ability of the proposedtechniques to compare heterogeneity of different datasets, and characterizeconvergence of the training process that deploys the introduced clientselection method. Extensive experimental results demonstrate that in non-IIDsettings HiCS-FL achieves faster convergence and lower training variance thanstate-of-the-art FL client selection schemes. Notably, HiCS-FL drasticallyreduces computation cost compared to existing selection schemes and isadaptable to different heterogeneity scenarios.

Accelerating Non-IID Federated Learning via Heterogeneity-Guided Client Sampling