Neurons vs Weights Pruning in Artificial Neural Networks
Environment. Technology. Resources: Proceedings of the 10th International Scientific and Practical Conference. Volume 3 2015
Andrejs Bondarenko, Arkādijs Borisovs, Ludmila Aleksejeva

Artificial neural networks (ANN) are well known for their good classification abilities. Recent advances in deep learning imposed second ANN renaissance. But neural networks possesses some problems like choosing hyper parameters such as neuron layers count and sizes which can greatly influence classification rate. Thus pruning techniques were developed that can reduce network sizes, increase its generalization abilities and overcome overfitting. Pruning approaches, in contrast to growing neural networks approach, assume that sufficiently large ANN is already trained and can be simplified with acceptable classification accuracy loss. Current paper compares nodes vs weights pruning algorithms and gives experimental results for pruned networks accuracy rates versus their non-pruned counterparts. We conclude that nodes pruning is more preferable solution, with some sidenotes.


Atslēgas vārdi
artificial neural networks, generalization, overfitting, pruning
DOI
10.17770/etr2015vol3.166
Hipersaite
http://journals.ru.lv/index.php/ETR/article/view/166

Bondarenko, A., Borisovs, A., Aleksejeva, L. Neurons vs Weights Pruning in Artificial Neural Networks. No: Environment. Technology. Resources: Proceedings of the 10th International Scientific and Practical Conference. Volume 3, Latvija, Rēzekne, 18.-20. jūnijs, 2015. Rēzekne: Rēzeknes Augstskola, 2015, 22.-28.lpp. ISBN 978-9984-44-173-3. ISSN 1691-5402. e-ISSN 2256-070X. Pieejams: doi:10.17770/etr2015vol3.166

Publikācijas valoda
English (en)
RTU Zinātniskā bibliotēka.
E-pasts: uzzinas@rtu.lv; Tālr: +371 28399196