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

With the increasing demand to efficiently deploy DNNs on mobile edge devices,it becomes much more important to reduce unnecessary computation and increasethe execution speed. Prior methods towards this goal, including modelcompression and network architecture search (NAS), are largely performedindependently and do not fully consider compiler-level optimizations which is amust-do for mobile acceleration. In this work, we first propose (i) a generalcategory of fine-grained structured pruning applicable to various DNN layers,and (ii) a comprehensive, compiler automatic code generation frameworksupporting different DNNs and different pruning schemes, which bridge the gapof model compression and NAS. We further propose NPAS, a compiler-aware unifiednetwork pruning, and architecture search. To deal with large search space, wepropose a meta-modeling procedure based on reinforcement learning with fastevaluation and Bayesian optimization, ensuring the total number of trainingepochs comparable with representative NAS frameworks. Our framework achieves6.7ms, 5.9ms, 3.9ms ImageNet inference times with 78.2%, 75% (MobileNet-V3level), and 71% (MobileNet-V2 level) Top-1 accuracy respectively on anoff-the-shelf mobile phone, consistently outperforming prior work.

NPAS: A Compiler-aware Framework of Unified Network Pruning and Architecture Search for Beyond Real-Time Mobile Acceleration