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

During the deployment of deep neural networks (DNNs) on edge devices, manyresearch efforts are devoted to the limited hardware resource. However, littleattention is paid to the influence of dynamic power management. As edge devicestypically only have a budget of energy with batteries (rather than almostunlimited energy support on servers or workstations), their dynamic powermanagement often changes the execution frequency as in the widely-used dynamicvoltage and frequency scaling (DVFS) technique. This leads to highly unstableinference speed performance, especially for computation-intensive DNN models,which can harm user experience and waste hardware resources. We firstlyidentify this problem and then propose All-in-One, a highly representativepruning framework to work with dynamic power management using DVFS. Theframework can use only one set of model weights and soft masks (together withother auxiliary parameters of negligible storage) to represent multiple modelsof various pruning ratios. By re-configuring the model to the correspondingpruning ratio for a specific execution frequency (and voltage), we are able toachieve stable inference speed, i.e., keeping the difference in speedperformance under various execution frequencies as small as possible. Ourexperiments demonstrate that our method not only achieves high accuracy formultiple models of different pruning ratios, but also reduces their variance ofinference latency for various frequencies, with minimal memory consumption ofonly one model and one soft mask.

All-in-One: A Highly Representative DNN Pruning Framework for Edge Devices with Dynamic Power Management