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

There exists a plethora of techniques for inducing structured sparsity inparametric models during the optimization process, with the final goal ofresource-efficient inference. However, few methods target a specific number offloating-point operations (FLOPs) as part of the optimization objective,despite many reporting FLOPs as part of the results. Furthermore, aone-size-fits-all approach ignores realistic system constraints, which differsignificantly between, say, a GPU and a mobile phone -- FLOPs on the formerincur less latency than on the latter; thus, it is important for practitionersto be able to specify a target number of FLOPs during model compression. Inthis work, we extend a state-of-the-art technique to directly incorporate FLOPsas part of the optimization objective and show that, given a desired FLOPsrequirement, different neural networks can be successfully trained for imageclassification.

FLOPs as a Direct Optimization Objective for Learning Sparse Neural Networks