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

To obtain excellent deep neural architectures, a series of techniques arecarefully designed in EfficientNets. The giant formula for simultaneouslyenlarging the resolution, depth and width provides us a Rubik's cube for neuralnetworks. So that we can find networks with high efficiency and excellentperformance by twisting the three dimensions. This paper aims to explore thetwisting rules for obtaining deep neural networks with minimum model sizes andcomputational costs. Different from the network enlarging, we observe thatresolution and depth are more important than width for tiny networks.Therefore, the original method, i.e., the compound scaling in EfficientNet isno longer suitable. To this end, we summarize a tiny formula for downsizingneural architectures through a series of smaller models derived from theEfficientNet-B0 with the FLOPs constraint. Experimental results on the ImageNetbenchmark illustrate that our TinyNet performs much better than the smallerversion of EfficientNets using the inversed giant formula. For instance, ourTinyNet-E achieves a 59.9% Top-1 accuracy with only 24M FLOPs, which is about1.9% higher than that of the previous best MobileNetV3 with similarcomputational cost. Code will be available athttps://github.com/huawei-noah/ghostnet/tree/master/tinynet_pytorch, andhttps://gitee.com/mindspore/mindspore/tree/master/model_zoo/research/cv/tinynet.

Model Rubik's Cube: Twisting Resolution, Depth and Width for TinyNets