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

Deep Neural Networks (DNNs) have begun to thrive in the field of automationsystems, owing to the recent advancements in standardising various aspects suchas architecture, optimization techniques, and regularization. In this paper, wetake a step towards a better understanding of Spectral Normalization (SN) andits potential for standardizing regularization of a wider range of DeepLearning models, following an empirical approach. We conduct severalexperiments to study their training dynamics, in comparison with the ubiquitousBatch Normalization (BN) and show that SN increases the gradient sparsity andcontrols the gradient variance. Furthermore, we show that SN suffers from aphenomenon, we call the mean-drift effect, which mitigates its performance. We,then, propose a weight reparameterization called as the Mean SpectralNormalization (MSN) to resolve the mean drift, thereby significantly improvingthe network's performance. Our model performs ~16% faster as compared to BN inpractice, and has fewer trainable parameters. We also show the performance ofour MSN for small, medium, and large CNNs - 3-layer CNN, VGG7 and DenseNet-BC,respectively - and unsupervised image generation tasks using GenerativeAdversarial Networks (GANs) to evaluate its applicability for a broad range ofembedded automation tasks.

Mean Spectral Normalization of Deep Neural Networks for Embedded Automation