09:16:05	 From Anna S  To  Troels Christian Petersen(privately) : You're breaking up. maybe turn off video?
09:23:36	 From Peter Andresen  To  Troels Christian Petersen(privately) : But do the numbers make sense without uncertainties ;) ?
09:31:49	 From Jakob Riber Rasmussen : Do you know
09:32:00	 From Jakob Riber Rasmussen : of any ML methods for smaller sample sizes?
09:33:24	 From Jakob Riber Rasmussen : Thanks
09:46:04	 From Norman Pedersen : How did you find out the amount of empty bins ?
09:49:01	 From Jakob Riber Rasmussen : One question regarding the binning.
09:50:24	 From Amalie Albrechtsen : Will you be going through the problem set at some point?
09:51:16	 From Jakob Riber Rasmussen : yeah thanks
09:57:32	 From Laurits Høgel : Bad connection
09:58:44	 From René : The latest state-of-the-art GAN is now StyleGAN2-ada and it is very cool! and open source by Nvidia: https://github.com/NVlabs/stylegan2-ada Paper  https://arxiv.org/abs/2006.06676. 
10:00:47	 From Norman Pedersen : Could you elaborate on how exactly it is penalized?
10:02:23	 From René : The full original loss function is in the original paper by Goodfellow https://arxiv.org/abs/1406.2661. It is basically a min max game between the generator and discriminator.