Generative Adversarial Networks (GANs) Simplified & Visualized | GANLAB
In this tutorial, we will explain the basics of GANS or generative Adversarial Networks.
We are going to use GANLab which is an incredible tool used to visualize GANS.
https://poloclub.github.io/ganlab/
“GANs are the most interesting idea in the last 10 years in Machine Learning”, Yann LeCun
GANs were developed by Ian GoodFellow in 2014 and are formed of two competing networks known as the generator and discriminator.
GANs work by having a generator network (counterfeiter) who is being trained to create fake dollars that are indistinguishable from the real ones (generated by the bank).
The discriminator network (police) is being trained to determine if the money is real or fake.
The counterfeiter is trying to fool the police by pretending that he generated a real dollar bill.
But, the discriminator will detect the fake money and provide feedback to the generator on why does he think that the money is fake.
Overtime, the generator will become expert in generating new money that are indistinguishable from the real ones and the discriminator will fail to tell the difference.
GANs are capable of generating new images that have never existed before.
GANs learn about the features of objects and create their own images.
GANs consists of a generator and discriminator.
Both the generator and discriminator start from scratch and learn together.
The Generator will generate images and the discriminator will compare these newly generated images (fake ones) to the true images (real ones) that are contained within the training dataset.
The generator works by trying to fool the discriminator by convincing it that the newly generated images are real ones.
Both networks learn together until the generator becomes a master at generating brand new images that are indistinguishable from the real ones.
I hope you have found the video useful and informative!
Happy Learning
#GAN #GenerativeAdversarialNetworks #GANLAB #DeepLearning
In this tutorial, we will explain the basics of GANS or generative Adversarial Networks.
We are going to use GANLab which is an incredible tool used to visualize GANS.
https://poloclub.github.io/ganlab/
“GANs are the most interesting idea in the last 10 years in Machine Learning”, Yann LeCun
GANs were developed by Ian GoodFellow in 2014 and are formed of two competing networks known as the generator and discriminator.
GANs work by having a generator network (counterfeiter) who is being trained to create fake dollars that are indistinguishable from the real ones (generated by the bank).
The discriminator network (police) is being trained to determine if the money is real or fake.
The counterfeiter is trying to fool the police by pretending that he generated a real dollar bill.
But, the discriminator will detect the fake money and provide feedback to the generator on why does he think that the money is fake.
Overtime, the generator will become expert in generating new money that are indistinguishable from the real ones and the discriminator will fail to tell the difference.
GANs are capable of generating new images that have never existed before.
GANs learn about the features of objects and create their own images.
GANs consists of a generator and discriminator.
Both the generator and discriminator start from scratch and learn together.
The Generator will generate images and the discriminator will compare these newly generated images (fake ones) to the true images (real ones) that are contained within the training dataset.
The generator works by trying to fool the discriminator by convincing it that the newly generated images are real ones.
Both networks learn together until the generator becomes a master at generating brand new images that are indistinguishable from the real ones.
I hope you have found the video useful and informative!
Happy Learning
#GAN #GenerativeAdversarialNetworks #GANLAB #DeepLearning
