Corpus ID: 52284330

Unconstrained Control of Feature Map Size Using Non-integer Strided Sampling

@inproceedings{Joo2018UnconstrainedCO,
  title={Unconstrained Control of Feature Map Size Using Non-integer Strided Sampling},
  author={Donggyu Joo and Junho Yim and Junmo Kim},
  booktitle={BMVC},
  year={2018}
}
Convolutional neural network (CNN) is mainly composed of convolution, pooling, and non-linear activation layers. Nowadays, almost all networks use only 2×2 max pooling or convolution layers with stride of 2 for downsampling. This technique is known to be good at extracting good feature, but it also has the constraint that feature map size is always reduced dramatically to half. In this work, we propose a simple new sampling technique that we call non-integer strided sampling (NSS), which… Expand

References

SHOWING 1-10 OF 22 REFERENCES
Deep Pyramidal Residual Networks
Fractional Max-Pooling
Densely Connected Convolutional Networks
Very Deep Convolutional Networks for Large-Scale Image Recognition
Deep Networks with Stochastic Depth
Going deeper with convolutions
Deep Residual Learning for Image Recognition
Learning Multiple Layers of Features from Tiny Images
Visualizing and Understanding Convolutional Networks
...
1
2
3
...