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

Federated learning (FL) is a promising paradigm that enables collaborativelylearning a shared model across massive clients while keeping the training datalocally. However, for many existing FL systems, clients need to frequentlyexchange model parameters of large data size with the remote cloud serverdirectly via wide-area networks (WAN), leading to significant communicationoverhead and long transmission time. To mitigate the communication bottleneck,we resort to the hierarchical federated learning paradigm of HiFL, which reapsthe benefits of mobile edge computing and combines synchronous client-edgemodel aggregation and asynchronous edge-cloud model aggregation together togreatly reduce the traffic volumes of WAN transmissions. Specifically, we firstanalyze the convergence bound of HiFL theoretically and identify the keycontrollable factors for model performance improvement. We then advocate anenhanced design of HiFlash by innovatively integrating deep reinforcementlearning based adaptive staleness control and heterogeneity-aware client-edgeassociation strategy to boost the system efficiency and mitigate the stalenesseffect without compromising model accuracy. Extensive experiments corroboratethe superior performance of HiFlash in model accuracy, communication reduction,and system efficiency.

HiFlash: Communication-Efficient Hierarchical Federated Learning with Adaptive Staleness Control and Heterogeneity-aware Client-Edge Association