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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Recent advances in artificial intelligence have driven increasing intelligentapplications at the network edge, such as smart home, smart factory, and smartcity. To deploy computationally intensive Deep Neural Networks (DNNs) onresource-constrained edge devices, traditional approaches have relied on eitheroffloading workload to the remote cloud or optimizing computation at the enddevice locally. However, the cloud-assisted approaches suffer from theunreliable and delay-significant wide-area network, and the local computingapproaches are limited by the constrained computing capability. Towardshigh-performance edge intelligence, the cooperative execution mechanism offersa new paradigm, which has attracted growing research interest recently. In thispaper, we propose CoEdge, a distributed DNN computing system that orchestratescooperative DNN inference over heterogeneous edge devices. CoEdge utilizesavailable computation and communication resources at the edge and dynamicallypartitions the DNN inference workload adaptive to devices' computingcapabilities and network conditions. Experimental evaluations based on arealistic prototype show that CoEdge outperforms status-quo approaches insaving energy with close inference latency, achieving up to 25.5%~66.9% energyreduction for four widely-adopted CNN models.

CoEdge: Cooperative DNN Inference with Adaptive Workload Partitioning over Heterogeneous Edge Devices