[
https://issues.apache.org/jira/browse/SYSTEMML-1483?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15995940#comment-15995940
]
Prithviraj Sen commented on SYSTEMML-1483:
------------------------------------------
Do you mean with no padding and stride=1? Under those assumptions,
conv2d_transpose with a 3X3 input and a 2X2 filter should return a 4X4 output
(equivalent to running a direct 'full' convolution).
I checked again. I think my args are in the right order. My filter is the first
arg to conv2d_backward_data and delta map the second arg.
> Add Deconvolution layer in nn library and Caffe2DML
> ---------------------------------------------------
>
> Key: SYSTEMML-1483
> URL: https://issues.apache.org/jira/browse/SYSTEMML-1483
> Project: SystemML
> Issue Type: Sub-task
> Reporter: Niketan Pansare
>
> http://caffe.berkeleyvision.org/tutorial/layers/deconvolution.html
> [~mwdus...@us.ibm.com] [~prithvi_r_s]
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
