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

Collaborative inference systems are one of the emerging solutions fordeploying deep neural networks (DNNs) at the wireless network edge. Their mainidea is to divide a DNN into two parts, where the first is shallow enough to bereliably executed at edge devices of limited computational power, while thesecond part is executed at an edge server with higher computationalcapabilities. The main advantage of such systems is that the input of the DNNgets compressed as the subsequent layers of the shallow part extract only theinformation necessary for the task. As a result, significant communicationsavings can be achieved compared to transmitting raw input samples. In thiswork, we study early exiting in the context of collaborative inference, whichallows obtaining inference results at the edge device for certain samples,without the need to transmit the partially processed data to the edge server atall, leading to further communication savings. The central part of our systemis the transmission-decision (TD) mechanism, which, given the information fromthe early exit, and the wireless channel conditions, decides whether to keepthe early exit prediction or transmit the data to the edge server for furtherprocessing. In this paper, we evaluate various TD mechanisms and showexperimentally, that for an image classification task over the wireless edge,proper utilization of early exits can provide both performance gains andsignificant communication savings.

Adaptive Early Exiting for Collaborative Inference over Noisy Wireless Channels