Live data from Hacker News

This Cat Does Not Exist

thiscatdoesnotexist.com

181–189 of 189 posts

Re: This Cat Does Not Exist

#183
post #55

I think this might be a well-done satire showcasing the pitfalls of having insufficient training data. It's magnificent. Half the cats are adorable and half are, I believe, creatures from "The Thing".

I suspect that the problem is less one of insufficient training data, and more one of excessively noisy training data.

There are three of these generators that have shown up on HN in the last few days—people, cats, and anime faces—and what the other two (more successful) ones have in common is that the things they're trying to generate all have the same basic shape and structure: that of a face.

The cat images in the training data are clearly just cats from every angle. There's much less of a clear structure for the neural net to recognize and reproduce.

That it's doing as well as it is is actually kind of remarkable, given that issue.

Re: This Cat Does Not Exist

#185
post #56

Also this airbnb doesnt exist https://thisairbnbdoesnotexist.com/ and these recordings dont really exist https://www.youtube.com/watch?v=llG-jQf8IBk&feature=youtu.be

And these Stack Overflow/Exchange questions don’t exist: https://se-simulator.lw1.at/ Disclaimer: Fun useless side project by me (And contains no machine learning, but just plain Markov chains)

you have no idea how similar this is to SO's moderation queue content

Re: This Cat Does Not Exist

#186
post #55

I think this might be a well-done satire showcasing the pitfalls of having insufficient training data. It's magnificent. Half the cats are adorable and half are, I believe, creatures from "The Thing".

Complete nightmare fuel. https://imgur.com/a/D4sRn5F

Tomorrow's Google reCaptcha

Re: This Cat Does Not Exist

#189
post #119

I still can't shake the feeling that most of these StyleGAN images are cleverly overfitting and just showing the face of an already existing cat in its training data. (But would love to be proven wrong!)

Back when I used to experiment with Markov chat simulators, this was a big problem. Besides the disappointment of finding out a particularly clever generated sentence was actually verbatim from the training set, there's also "accidental sharing" and/or "plagarism" issues. Of course with text it's pretty simple to code a check that output doesn't exactly match any known inputs. Not sure how you'd do that with images;…

Yes. For hysterical raisins, VGG-16 is usually used as the hash/space for the nearest-neighbor lookups. Recent example of this is in the BigGAN appendix, where you can see that despite the dog samples looking perfect, they are nevertheless totally different from their closest neighbors in the ImageNet training data and so can't be memorization.
Post reply on HN