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

Along with the popularity of Artificial Intelligence (AI) andInternet-of-Things (IoT), Federated Learning (FL) has attracted steadilyincreasing attentions as a promising distributed machine learning paradigm,which enables the training of a central model on for numerous decentralizeddevices without exposing their privacy. However, due to the biased datadistributions on involved devices, FL inherently suffers from lowclassification accuracy in non-IID scenarios. Although various device groupingmethod have been proposed to address this problem, most of them neglect both i)distinct data distribution characteristics of heterogeneous devices, and ii)contributions and hazards of local models, which are extremely important indetermining the quality of global model aggregation. In this paper, we presentan effective FL method named FedEntropy with a novel dynamic device groupingscheme, which makes full use of the above two factors based on our proposedmaximum entropy judgement heuristic.Unlike existing FL methods that directlyaggregate local models returned from all the selected devices, in one FL roundFedEntropy firstly makes a judgement based on the pre-collected soft labels ofselected devices and then only aggregates the local models that can maximizethe overall entropy of these soft labels. Without collecting local models thatare harmful for aggregation, FedEntropy can effectively improve global modelaccuracy while reducing the overall communication overhead. Comprehensiveexperimental results on well-known benchmarks show that, FedEntropy not onlyoutperforms state-of-the-art FL methods in terms of model accuracy andcommunication overhead, but also can be integrated into them to enhance theirclassification performance.

FedEntropy: Efficient Device Grouping for Federated Learning Using Maximum Entropy Judgment