Loading...
Loading...

Go to the content (press return)

FuCiTNet: improving the generalization of deep learning networks by the fusion of learned class-inherent transformations

Author
Rey-Arena, M.; Guirado, E.; Tabik, S.; Ruiz-Hidalgo, J.
Type of activity
Journal article
Journal
Information fusion
Date of publication
2020-10
Volume
63
First page
188
Last page
195
DOI
10.1016/j.inffus.2020.06.015
Project funding
MALEGRA, TEC2016-75976-R
Repository
http://hdl.handle.net/2117/328939 Open in new window
URL
https://www.sciencedirect.com/science/article/abs/pii/S1566253520303122 Open in new window
Abstract
It is widely known that very small datasets produce overfitting in Deep Neural Networks (DNNs), i.e., the network becomes highly biased to the data it has been trained on. This issue is often alleviated using transfer learning, regularization techniques and/or data augmentation. This work presents a new approach, independent but complementary to the previous mentioned techniques, for improving the generalization of DNNs on very small datasets in which the involved classes share many visual featu...
Citation
Rey-Arena, M. [et al.]. FuCiTNet: improving the generalization of deep learning networks by the fusion of learned class-inherent transformations. "Information fusion", Octubre 2020, vol. 63, p. 188-195.
Keywords
Classification, Deep neural networks, GANs (Generative Adversarial Networks), Generalization, Pre-processing, Small dataset, Transformation
Group of research
GPI - Image and Video Processing Group
IDEAI-UPC - Intelligent Data Science and Artificial Intelligence Research Center

Participants

  • Rey-Arena, Manuel  (author)
  • Guirado, Emilio  (author)
  • Tabik, Siham  (author)
  • Ruiz Hidalgo, Javier  (author)