It looks to me like Conv2DTranspose in the decoder replaces using Conv2D and UpSampling.

Building a Convolutional Autoencoder using Keras using Conv2DTranspose
426
4
Surajit Saikia
Evan Zamir
·Follow
1 min read·
Oct 14, 2019
--
It looks to me like Conv2DTranspose in the decoder replaces using Conv2D and UpSampling. Is that right? Are there advantages doing it one way or the other?
--
--
Written by Evan Zamir459 Followers
·148 Following
Data Scientist. San Francisco.
Responses (1)
Help
Status
About
Careers
Press
Blog
Privacy
Terms
Text to speech
Teams