5 ms·
Any Deepfake detection tool would also be a GAN sparring partner for the Deepfake makers.
by dzink 8y ago
Any Deepfake detection tool would also be a GAN sparring partner for the Deepfake makers.
- EGreg 8y agoIf you consider this a two sided game, then you can have the fakes quickly become so good that hardly anyone can tell. And I'm not just talking about image fakes. You can have MCTS find the best arguments for ludicrous statements, and other paths to justify fake things that look just like real arguments.
- ben_w 8y agoThat is the point of a GAN. We’re missing something about the problem space and/or human cognition or their output would already be indistinguishable from reality to all humans.
- varelse 8y agoYep, they're a 3D reconstruction of points from a 4D space a (massively simplifying the fact that the 2D video frames themselves are high-dimensional data) and that's just the video. Bring a 3D camera into the game and the underlying distribution and see if hilarity ensues IMO.
- amelius 8y agoHence they probably shouldn't make it public for it to be effective.
- musage 8y agoEffective how? The article quote someone talking about "saving democracy", but the use of the military saying "we think X is fake but we can't tell how we came to that conclusion" seems very limited, and depending on the content and target audience of the fake would just get a shrug and "of course they'd say that". Are there even past examples of any military, US or otherwise, injecting itself into public discourse over something to clear things up, instead of doing the usual opposite?
- some_random 8y agoBut then everyone has to trust that whatever the US Military says about the video is true, which kinda ruins the point.
- bmc7505 8y agoOr: they only make it public once the discriminator demonstrates a large margin to reduce the potential training signal.
- romaniv 8y agoThis doesn't automatically mean GANs can outpace any detection method.
- smallnamespace 8y agoIf the detection methods are publicly available, then you simply incorporate the detection method in your training regimen. Effective detection methods may end up being closely guarded secrets.
- GunwaleHamilton 8y agoWhich means we would have to rely on an "Authority" who cannot be independently verified, because we do not know the secret sauce they are using.
- londons_explore 8y agoThey need to be publicly available and differentiable to be able to use as an opponent for a GAN. Pretty much nothing except other neural networks are differentiable, unless you put effort into designing them to be.
- supermdguy 8y agoIt's feasible to design a reinforcement learning based network that uses output from a non-differentiable deepfake detector as a component of the loss function.
- mikec3010 8y agoAnd can you imagine the impact of false positives? Prosecutor has the perp on video committing a heinous offense but it's still insufficient evidence because of a software bug.
- jessaustin 8y agoWe can dream of juries that wise... haven't seen one yet though have we?