5,321
edits
Convolutional neural network: Difference between revisions
Convolutional neural network (view source)
Revision as of 17:26, 22 September 2020
, 22 September 2020→Dilation
(→Kernel) |
|||
| Line 115: | Line 115: | ||
Space between pixels in the kernel<br> | Space between pixels in the kernel<br> | ||
A dilation of 1 will apply a 3x3 kernel over a 5x5 region. This would be equivalent to a 5x5 kernel with odd index weights (<math> i \% 2 == 1</math>) set to 0. | A dilation of 1 will apply a 3x3 kernel over a 5x5 region. This would be equivalent to a 5x5 kernel with odd index weights (<math> i \% 2 == 1</math>) set to 0. | ||
* These are used in Atrous Spatial Pyramid Pooling (ASPP), where different dilation amounts mimic a UNet's spatial pyramid. | |||
** The pro is that ASPP uses less computation since there are fewer kernels (i.e. fewer channels). The con is that it takes up more memory. | |||
===Groups=== | ===Groups=== | ||