They're abusing a dumber nearest neighbor algorithm though, which is easy. I had to abuse PNG metadata to make it work with scalers that at least attempt to have a modicum of quality suitable for humans. Why are we training neural networks on images with horrible aliasing?
-
-
Show this threadThanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
can't believe old 4chan memes are what passes for an "attack" these days
-
I'm not convinced they've used any ML besides resnet tbh. Like, 'can we trust self-driving cars and facial recognition' makes no sense when any decent ML in those categories would probably be using generated data anyhow. At most this affects social media tagging.
- Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.