Non-linear Convolution Filters for CNN-Based Learning

@article{Zoumpourlis2017NonlinearCF,
  title={Non-linear Convolution Filters for CNN-Based Learning},
  author={Georgios Zoumpourlis and Alexandros Doumanoglou and N. Vretos and P. Daras},
  journal={2017 IEEE International Conference on Computer Vision (ICCV)},
  year={2017},
  pages={4771-4779}
}
During the last years, Convolutional Neural Networks (CNNs) have achieved state-of-the-art performance in image classification. [...] Key Method Such forms, constituting a more rich function space, are used as approximations of the response profile of visual cells. Our proposed second-order convolution is tested on CIFAR-10 and CIFAR-100. We show that a network which combines linear and non-linear filters in its convolutional layers, can outperform networks that use standard linear filters with the same…Expand
39 Citations
Improving CNN linear layers with power mean non-linearity
  • 9
  • PDF
Kervolutional Neural Networks
  • 27
  • Highly Influenced
  • PDF
Operational vs Convolutional Neural Networks for Image Denoising
  • PDF
IC Networks: Remodeling the Basic Unit for Convolutional Neural Networks
  • PDF
Global Second-Order Pooling Convolutional Networks
  • 49
  • PDF
A Push-Pull Layer Improves Robustness of Convolutional Neural Networks
  • 2
  • PDF
Learning the Non-linearity in Convolutional Neural Networks
  • 1
  • PDF
Enhanced robustness of convolutional networks with a push–pull inhibition layer
  • 5
  • PDF
IC-Network: Efficient Structure for Convolutional Neural Networks
  • PDF
...
1
2
3
4
...

References

SHOWING 1-10 OF 38 REFERENCES
Recurrent convolutional neural network for object recognition
  • Ming Liang, Xiaolin Hu
  • Computer Science
  • 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
  • 2015
  • 628
  • PDF
Network In Network
  • 3,348
  • PDF
Estimating nonlinear receptive fields from natural images.
  • 25
  • PDF
Deep Pyramidal Residual Networks
  • 303
  • PDF
Learning Multiple Layers of Features from Tiny Images
  • 10,897
  • Highly Influential
  • PDF
Oriented Response Networks
  • 132
  • PDF
Deeply-Supervised Nets
  • 1,185
  • PDF
Going deeper with convolutions
  • 22,787
  • PDF
High Performance Convolutional Neural Networks for Document Processing
  • 345
  • PDF
...
1
2
3
4
...