Researchers who might be interested in this: @__ishaan @dumoulinv @goodfellow_ian @martinarjovsky @soumithchintala
-
-
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
I would also like to thank
@BorealisAI who helped me through their fellowship. Make sure to apply to this year Fellowship https://www.borealisai.com/en/careers/fellowships/ …, they extended their deadline to October 20.Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
It looks like you have put in 3 years of work in this. Experts might differ but I feel that content wise, it's a lot to think about in this paper.
-
Thank you for the compliment! I worked very hard on this. I try to generalize things as much as possible and present the big picture.
Kraj razgovora
Novi razgovor -
-
-
Hi
@jm_alexia, I liked the paper and it provided a new way to look at things. I was just wondering isnt "maximizing geometric margin a key to effective gan training" more because of using FID (norm based). What do you think would have happened if the metric was based on entropy? -
I don't see the link, but if you think that this could be true, it could make a nice paper. However, gradient penalties do work well even visually speaking (see StyleGAN and Progressive GAN results). FID really correlates well with good visuals and diversity.
Kraj razgovora
Novi razgovor -
-
-
I can't understand most of the paper or the blog post :( I definitely need to learn more math
-
Try reading the background section and chapter 7 of http://users.isr.ist.utl.pt/~wurmd/Livros/school/Bishop%20-%20Pattern%20Recognition%20And%20Machine%20Learning%20-%20Springer%20%202006.pdf ….
- Još 1 odgovor
Novi razgovor -
-
-
You might be interested in this old paper (was reminded of this paper after reading your abstract): https://www.ml.uni-saarland.de/Publications/HeiBouSch2005.pdf …
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
Appreciate the readable blog paper summary. Too bad emojis aren't standard in papers.pic.twitter.com/nfvtVJYNkO
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.
cats
, video games
and GANs
.
