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 this paper, we tackle the problem of convolutional neural network design.Instead of focusing on the design of the overall architecture, we investigate adesign space that is usually overlooked, i.e. adjusting the channelconfigurations of predefined networks. We find that this adjustment can beachieved by shrinking widened baseline networks and leads to superiorperformance. Based on that, we articulate the heterogeneity hypothesis: withthe same training protocol, there exists a layer-wise differentiated networkarchitecture (LW-DNA) that can outperform the original network with regularchannel configurations but with a lower level of model complexity.  The LW-DNA models are identified without extra computational cost or trainingtime compared with the original network. This constraint leads to controlledexperiments which direct the focus to the importance of layer-wise specificchannel configurations. LW-DNA models come with advantages related tooverfitting, i.e. the relative relationship between model complexity anddataset size. Experiments are conducted on various networks and datasets forimage classification, visual tracking and image restoration. The resultantLW-DNA models consistently outperform the baseline models. Code is available athttps://github.com/ofsoundof/Heterogeneity_Hypothesis.

The Heterogeneity Hypothesis: Finding Layer-Wise Differentiated Network Architectures