Pokemon GAN

How It Works

Pokemon GAN

How It Works

Generative adversarial networks, or GANs, are neural networks that can create completely new things, such as text (like books), pictures, and even music! Based on a huge amount of data, GANs can create realistic new things indisinguishable from what a human could create!

I recently created my own GAN. Check out the video I made here!

GANs are letting AI be "creative", but music and art aren't the only applications of GANs. They are being used in things like drug discovery, to cure previously incurable diseases like cancer, and even creating synthetic data, which is useful everywhere!

How it Works

Two Neural Networks

A GAN is made up of two neural networks. The generator creates new data based on what it is given, while the discriminator tries to tell the difference between real and fake data. This works for art, music, and even Pokemon!

Training

Based on a lot of data, the generator tries to find common features in the data and create new data with these features. The discriminator tries to find common features in the same data, and looks at real and fake data to determine which is which.

Applications

GANs have so many applications, and this isn't limited to Pokemon. They can be used to discover new drugs, create synthetic data (so you can train even more GANs), generate images (for those of us that want to avoid looking for high-quality images), and more!

Contact Me