Wasserstein GAN
Okay kiddo, let me explain Wasserstein GAN to you.
So imagine you are playing a game with your friend, and you have to draw a picture of an animal. Your friend gets to choose the animal, and they will tell you how good your drawing is. If your drawing is really good, they might give you a prize.
Now, the same thing happens when you use a Wasserstein GAN. Instead of drawing pictures of animals, you are trying to create realistic images using a computer program. The program is called a generator, and it makes images.
But there is someone else in the game, called the discriminator. This is like your friend who judges your animal drawings. The discriminator looks at the images that the generator makes, and decides if they are good or not. If the images are good, the discriminator gives the generator a prize, like a higher score.
But the generator doesn't want to just make any image - it wants to make images that look like they belong in real life. So it tries to make images that fool the discriminator into thinking they are real. The better the generator is at making realistic images, the more prizes it gets.
This is how a Wasserstein GAN works - it uses a special type of scoring system to judge how good the generator is at making realistic images. The scoring system is called a Wasserstein distance, and it is used to measure how different the images made by the generator are from real-life images.
Overall, a Wasserstein GAN is a way for a computer program to learn how to make realistic images, by playing a game with a judge and trying to win prizes. Cool, huh?
So imagine you are playing a game with your friend, and you have to draw a picture of an animal. Your friend gets to choose the animal, and they will tell you how good your drawing is. If your drawing is really good, they might give you a prize.
Now, the same thing happens when you use a Wasserstein GAN. Instead of drawing pictures of animals, you are trying to create realistic images using a computer program. The program is called a generator, and it makes images.
But there is someone else in the game, called the discriminator. This is like your friend who judges your animal drawings. The discriminator looks at the images that the generator makes, and decides if they are good or not. If the images are good, the discriminator gives the generator a prize, like a higher score.
But the generator doesn't want to just make any image - it wants to make images that look like they belong in real life. So it tries to make images that fool the discriminator into thinking they are real. The better the generator is at making realistic images, the more prizes it gets.
This is how a Wasserstein GAN works - it uses a special type of scoring system to judge how good the generator is at making realistic images. The scoring system is called a Wasserstein distance, and it is used to measure how different the images made by the generator are from real-life images.
Overall, a Wasserstein GAN is a way for a computer program to learn how to make realistic images, by playing a game with a judge and trying to win prizes. Cool, huh?