I'm applying conv2d_grad_wrt_inputs into one feature maps (the second image below) using its kernel. would interpret the third image below as a correct deconvolution since I'm seeing it as repetition of the some feature maps into the input.
<https://lh3.googleusercontent.com/-6c7HTwi--Lo/WGQR-YDndlI/AAAAAAAAA2U/V00_f_flHB0HNVtY1xpmnBCRZSS6ZTl8ACKgB/s1600/deconv_5.png> -- --- You received this message because you are subscribed to the Google Groups "theano-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
