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

Neuroevolution has greatly promoted Deep Neural Network (DNN) architecturedesign and its applications, while there is a lack of methods available acrossdifferent DNN types concerning both their scale and performance. In this study,we propose a self-adaptive neuroevolution (SANE) approach to automaticallyconstruct various lightweight DNN architectures for different tasks. One of thekey settings in SANE is the search space defined by cells and organsself-adapted to different DNN types. Based on this search space, a constructiveevolution strategy with uniform evolution settings and operations is designedto grow DNN architectures gradually. SANE is able to self-adaptively adjustevolution exploration and exploitation to improve search efficiency. Moreover,a speciation scheme is developed to protect evolution from early convergence byrestricting selection competition within species. To evaluate SANE, we carryout neuroevolution experiments to generate different DNN architecturesincluding convolutional neural network, generative adversarial network and longshort-term memory. The results illustrate that the obtained DNN architecturescould have smaller scale with similar performance compared to existing DNNarchitectures. Our proposed SANE provides an efficient approach toself-adaptively search DNN architectures across different types.

A Self-adaptive Neuroevolution Approach to Constructing Deep Neural Network Architectures Across Different Types