I just recently found your course and it has been amazing, many thanks :).
I’m working on a project that intends to autocomplete missing parts of 3d objects. These are broken archaeology objects that we scan from a museum (Hi from Perú ).
I’ve been working on a 3d autoencoder with a dropout at the very beginning (to simulate missing points), and now I’m reading/learning more about GANS to apply them in this work.
I would really appreciate if you have some ideas I can apply
Also, i would like to hear what do you think about this paper Improved Training of Wasserstein GANs ?