TRANSFER LEARNING FOR TRAINING ACCELARATION

Authors

  • Ilmārs Apeināns Rezekne Academy of Technologies (LV)
  • Vitālijs Žukovs Rezekne Academy of Technologies (LV)
  • Sergejs Kodors Rezekne Academy of Technologies (LV)
  • Imants Zarembo Rezekne Academy of Technologies (LV)

DOI:

https://doi.org/10.17770/het2021.25.6773

Keywords:

Alexnet, CNN, Convolution Neuron Network, Model, Transfer learning,

Abstract

In this work, authors compare training time of standard convolution neuron network model with model trained using transfer learning. Both models are based on Alexnet architecture. CNN model training from scratch included full model, but using transfer learning, some layers of model were frozen for learning acceleration considering transfer learning methodology.

Downloads

Download data is not yet available.

References

Konvolucionāli neironu tīkli vizuālai atpazīšanai [tiešsaiste], [atsauce uz 07.08.2020.]. Pieejams: https://cs231n.github.io/convolutional-networks/

Transfer learning pamatideja [tiešsaiste], [atsauce uz 18.08.2020.]. Pieejams: https://towardsdatascience.com/transfer-learning-with-convolutional-neural-networks-in-pytorch-dd09190245ce

Alexnet arhitektūra [tiešsaiste], [atsauce uz 12.08.2020.]. Pieejams: https://medium.com/@smallfishbigsea/a-walk-through-of-alexnet-6cbd137a5637

H. Mureșan & O. Mihai. Fruit recognition from images using deep learning. Acta Universitatis Sapientiae, Informatica, vol. 10, pp. 26-42, 2018.

CIFAR100 datu kopa [tiešsaiste], [atsauce uz 12.08.2020.].

Downloads

Published

2021-04-23

Issue

No. 25 (2021): Human. Environment. Technology

Section

Information Technologies

How to Cite

[1]
I. Apeināns, V. Žukovs, S. Kodors, and I. Zarembo, “TRANSFER LEARNING FOR TRAINING ACCELARATION”, HET, no. 25, pp. 16–21, Apr. 2021, doi: 10.17770/het2021.25.6773.