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

The deployment of Deep Neural Networks (DNNs) on edge devices is hindered bythe substantial gap between performance requirements and available processingpower. While recent research has made significant strides in developing pruningmethods to build a sparse network for reducing the computing overhead of DNNs,there remains considerable accuracy loss, especially at high pruning ratios. Wefind that the architectures designed for dense networks by differentiablearchitecture search methods are ineffective when pruning mechanisms are appliedto them. The main reason is that the current method does not support sparsearchitectures in their search space and uses a search objective that is madefor dense networks and does not pay any attention to sparsity. In this paper,we propose a new method to search for sparsity-friendly neural architectures.We do this by adding two new sparse operations to the search space andmodifying the search objective. We propose two novel parametric SparseConv andSparseLinear operations in order to expand the search space to include sparseoperations. In particular, these operations make a flexible search space due tousing sparse parametric versions of linear and convolution operations. Theproposed search objective lets us train the architecture based on the sparsityof the search space operations. Quantitative analyses demonstrate that oursearch architectures outperform those used in the stateof-the-art sparsenetworks on the CIFAR-10 and ImageNet datasets. In terms of performance andhardware effectiveness, DASS increases the accuracy of the sparse version ofMobileNet-v2 from 73.44% to 81.35% (+7.91% improvement) with 3.87x fasterinference time.

DASS: Differentiable Architecture Search for Sparse neural networks