@@nancyjones7932 It's to use a discriminator, which is what you already know as a classifier to train a generator, something that can create data, for example something that can create images. The better the discriminator the better the generator. One very important implication of this is that if you've made a model / algorithm to detect deepfakes, then this model / algorithm can be used to improve a generator to the point that your discriminator won't be able to tell the difference. Conversely, if you have a very good generator, you can use it to rapidly create a labeled dataset which you can use to train a discriminator / classifier.
![[Classic] Generative Adversarial Networks (Paper Explained)](http://i.ytimg.com/vi/eyxmSmjmNS0/mqdefault.jpg)
Man you explain things so freaking sharply, in ways no one thinks of - this should have at least 500k views
Thank you, this was one of the most intuitive understanding of GANs.
The explanation is outstanding. Thanks a lot!
wonderful GAN intro
Awesome explanation Dude
“A neural network can learn any function, including the probability distribution of images of faces.” Woah
Hi, can you also make a video on the Universal Approximation Theorem?
Great
Damn this shit is cool
Sir can we use gan for speech enhancement
Nice intro! Do you have a second part?
Thanks! And yes: Understand the Math and Theory of GANs in ~ 10 minutes th-cam.com/video/J1aG12dLo4I/w-d-xo.html
What is the purpose of creating GANs?
@@nancyjones7932 It's to use a discriminator, which is what you already know as a classifier to train a generator, something that can create data, for example something that can create images. The better the discriminator the better the generator.
One very important implication of this is that if you've made a model / algorithm to detect deepfakes, then this model / algorithm can be used to improve a generator to the point that your discriminator won't be able to tell the difference.
Conversely, if you have a very good generator, you can use it to rapidly create a labeled dataset which you can use to train a discriminator / classifier.