Intro to Generative Adversarial Networks (GANs)
Yesterday I gave a interactive workshop on Generative Adversarial Networks. These are a type of neural network that are used to generate new information that did not previously exist. They have been used from everything from creating photos of people that don't exist, to predicting stock prices.
One area that has been shown quite a lot recently is using GANs to automatically colour in black and white images. Such as using tools like DeOldify.
But how do they work? In this workshop I explain the basics of how GANs work, and the components that make up a GAN. I also walk through the code of a simple GAN built using PyTorch.
The full video of the workshop is below:
Or you can follow along in the interactive Python notebook available in the IBM Developer UK Github repository:
https://github.com/IBMDeveloperUK/Colourise-GAN-Workshop
If you want to have a go at running the GAN yourself the instructions are there in the readme file on how to get started with IBM Watson Studio and running the code.
You will need a free IBM Cloud account, which you can sign up for here.
If you want to stay in touch with our series of free events (wherever you are in the world right now), then join up to our meetup group: