4 ms·
Couldn't you easily infer the attacking noise by comparing the original and the changed images? Once you have the attacking noise it would be pretty trivial to
by endisneigh 6y ago
Couldn't you easily infer the attacking noise by comparing the original and the changed images? Once you have the attacking noise it would be pretty trivial to beat this, no?
I also don't see how this would do much against object recognition or face recognition. More insight to the types of recognition this actually fights against would be helpful.
- alexchamberlain 6y agoI think the idea is that the AI doesn't have access to the original. That being said, I'm not sure what would stop such AIs from being _trained_ on images that have been attacked.
- capableweb 6y agoWe'll end up in a similar cat&mouse game as online "pirates" been at for a long time. Developers create something to break the AI, the AI adopt because they figure out the noise profile, developers change the noise profile and the AI has to adopt again.
- xondono 6y ago> I also don't see how this would do much against object recognition or face recognition. That’s precisely the point, you are creating noise that humans are insensitive to, but that severely affects AI. The idea as I understand is that if you need to upload an image (of yourself for instance), you can use this to complicate matters to AIs by uploading the modified picture.