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) facilitates edge devices to cooperatively train aglobal shared model while maintaining the training data locally and privately.However, a common but impractical assumption in FL is that the participatingedge devices possess the same required resources and share identical globalmodel architecture. In this study, we propose a novel FL method calledFederated Intermediate Layers Learning (FedIN), supporting heterogeneous modelswithout utilizing any public dataset. The training models in FedIN are dividedinto three parts, including an extractor, the intermediate layers, and aclassifier. The model architectures of the extractor and classifier are thesame in all devices to maintain the consistency of the intermediate layerfeatures, while the architectures of the intermediate layers can vary forheterogeneous devices according to their resource capacities. To exploit theknowledge from features, we propose IN training, training the intermediatelayers in line with the features from other clients. Additionally, we formulateand solve a convex optimization problem to mitigate the gradient divergenceproblem induced by the conflicts between the IN training and the localtraining. The experiment results show that FedIN achieves the best performancein the heterogeneous model environment compared with the state-of-the-artalgorithms. Furthermore, our ablation study demonstrates the effectiveness ofIN training and the solution to the convex optimization problem.

FedIN: Federated Intermediate Layers Learning for Model Heterogeneity