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

Modern neural network modules which can significantly enhance the learningpower usually add too much computational complexity to the original neuralnetworks. In this paper, we pursue very efficient neural network modules whichcan significantly boost the learning power of deep convolutional neuralnetworks with negligible extra computational cost. We first present boththeoretically and experimentally that elementwise square operator has apotential to enhance the learning power of neural networks. Then, we designfour types of lightweight modules with elementwise square operators, named asSquare-Pooling, Square-Softmin, Square-Excitation, and Square-Encoding. We addour four lightweight modules to Resnet18, Resnet50, and ShuffleNetV2 for betterperformance in the experiment on ImageNet 2012 dataset. The experimentalresults show that our modules can bring significant accuracy improvements tothe base convolutional neural network models. The performance of ourlightweight modules is even comparable to many complicated modules such asbilinear pooling, Squeeze-and-Excitation, and Gather-Excite. Our highlyefficient modules are particularly suitable for mobile models. For example,when equipped with a single Square-Pooling module, the top-1 classificationaccuracy of ShuffleNetV2-0.5x on ImageNet 2012 is absolutely improved by 1.45%with no additional parameters and negligible inference time overhead.

DeepSquare: Boosting the Learning Power of Deep Convolutional Neural Networks with Elementwise Square Operators