To my machine vision friends on mastodon, is there a way to visualize what each filter is *generating* in a GAN? Sort of like what it's done in CNNs that gives us images like below. I found two papers doing some interesting things by e.g., removing filters that add trees to images so the generated image without them is treeless. Still, you don’t get to see what a filter did for a generated image.
| Website | https://www.kaysonfakhar.com |
| https://twitter.com/kaysonfakhar | |
| Spotify | https://open.spotify.com/artist/4V9FIRrYQ0drSzZm9YK3sk?si=x6EolXjYSOKOOJ6uGO1Qow |
