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

This paper aims to accelerate the test-time computation of convolutionalneural networks (CNNs), especially very deep CNNs that have substantiallyimpacted the computer vision community. Unlike previous methods that aredesigned for approximating linear filters or linear responses, our method takesthe nonlinear units into account. We develop an effective solution to theresulting nonlinear optimization problem without the need of stochasticgradient descent (SGD). More importantly, while previous methods mainly focuson optimizing one or two layers, our nonlinear method enables an asymmetricreconstruction that reduces the rapidly accumulated error when multiple (e.g.,>=10) layers are approximated. For the widely used very deep VGG-16 model, ourmethod achieves a whole-model speedup of 4x with merely a 0.3% increase oftop-5 error in ImageNet classification. Our 4x accelerated VGG-16 model alsoshows a graceful accuracy degradation for object detection when plugged intothe Fast R-CNN detector.

Accelerating Very Deep Convolutional Networks for Classification and Detection