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

In order to deploy deep convolutional neural networks (CNNs) onresource-limited devices, many model pruning methods for filters and weightshave been developed, while only a few to layer pruning. However, compared withfilter pruning and weight pruning, the compact model obtained by layer pruninghas less inference time and run-time memory usage when the same FLOPs andnumber of parameters are pruned because of less data moving in memory. In thispaper, we propose a simple layer pruning method using fusible residualconvolutional block (ResConv), which is implemented by inserting shortcutconnection with a trainable information control parameter into a singleconvolutional layer. Using ResConv structures in training can improve networkaccuracy and train deep plain networks, and adds no additional computationduring inference process because ResConv is fused to be an ordinaryconvolutional layer after training. For layer pruning, we convert convolutionallayers of network into ResConv with a layer scaling factor. In the trainingprocess, the L1 regularization is adopted to make the scaling factors sparse,so that unimportant layers are automatically identified and then removed,resulting in a model of layer reduction. Our pruning method achieves excellentperformance of compression and acceleration over the state-of-the-arts ondifferent datasets, and needs no retraining in the case of low pruning rate.For example, with ResNet-110, we achieve a 65.5%-FLOPs reduction by removing55.5% of the parameters, with only a small loss of 0.13% in top-1 accuracy onCIFAR-10.

Layer Pruning via Fusible Residual Convolutional Block for Deep Neural Networks