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

Many recent studies integrate federated learning (FL) with self-supervisedlearning (SSL) to take advantage of raw training data distributed across edgedevices. However, edge devices often struggle with high computation andcommunication costs imposed by SSL and FL algorithms. To tackle this hindrance,we propose LW-FedSSL, a layer-wise federated self-supervised learning approachthat allows edge devices to incrementally train one layer of the model at atime. LW-FedSSL comprises server-side calibration and representation alignmentmechanisms to maintain comparable performance with end-to-end FedSSL whilesignificantly lowering clients' resource requirements. The server-sidecalibration mechanism takes advantage of the resource-rich server in an FLenvironment to assist in global model training. Meanwhile, the representationalignment mechanism encourages closeness between representations of FL localmodels and those of the global model. Our experiments show that LW-FedSSL has a$3.3 \times$ lower memory requirement and a $3.2 \times$ cheaper communicationcost than its end-to-end counterpart. We also explore a progressive trainingstrategy called Prog-FedSSL that outperforms end-to-end training with a similarmemory requirement and a $1.8 \times$ cheaper communication cost.

LW-FedSSL: Resource-efficient Layer-wise Federated Self-supervised Learning