I would like to know how 'SAME' padding is calculated when using a transpose convolutional layer in Tensorflow. I find the way it is calculated in this link in a convolutional layer with this formula in this link, but I wonder how it is done in the deconvolutional layer. Thanks!
Asked
Active
Viewed 301 times
2
-
There is an answer for a similar [question](https://stackoverflow.com/questions/50683039/conv2d-transpose-output-shape-using-formula) – Mohan Radhakrishnan Oct 04 '18 at 15:04