Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

That wasn't the point. The point was, contrary to your claims, CNNs are indeed all about statistical correlation.


It seems odd that GANs would learn to generate images with precise yet lifelike shape and texture based purely on statistical correlation.


No it doesn't. Because that's what the CNNs used by the GANs are fitting on, cross correlation.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: