deep-INFOMAX icon indicating copy to clipboard operation
deep-INFOMAX copied to clipboard

Published 20 hours ago • verified publisher icon rcalland

No striding for convolutions in the encoder

Open guillaumelorre28 opened this issue 6 years ago • 0 comments

In the article, they use a DCGAN architecture for the encoder, which uses a striding of 2 for every convolutions in the encoder.

guillaumelorre28 avatar Sep 28 '18 08:09 guillaumelorre28