4 ms·
GAN networks make this a flawed endeavor. you can train a network to detect it, then a network to convince it, ad nauseum. https://en.wikipedia.org/wiki/Gener
by Jerrrry 5y ago
GAN networks make this a flawed endeavor.
you can train a network to detect it, then a network to convince it, ad nauseum.
https://en.wikipedia.org/wiki/Generative_adversarial_network https://en.wikipedia.org/wiki/Generative_adversarial_network
- avodonosov 5y agoI was hoping some inevitable property of deep fakes, that could help detecting them.
- Jerrrry 5y agoany such property will eventually be selected against by the discriminator. https://developers.google.com/machine-learning/gan/discriminator https://developers.google.com/machine-learning/gan/discrimin...