Code Walkthrough for Breakthrough Wasserstein GAN:Build Your Own Image Generator

Join us to walk through the code for Wasserstein GAN (https://github.com/martinarjovsky/WassersteinGAN), including details about how to get the examples from the paper (https://github.com/martinarjovsky/WassersteinGAN#generator-sample-quality-correlates-with-discriminator-loss) running. People who are set up to use AWS will be able to run and play with their own instances of the code.

Wasserstein GAN is a new, very robust way to train an image generator from a set of example images (similar to InfoGAN, which we covered previously (https://www.meetup.com/Cambridge-Artificial-Intelligence-Meetup/events/236122160/).)

The paper (https://arxiv.org/pdf/1701.07875.pdf) involves some forbidding math, but the operation of the code can be explained in simple terms, and we will do so.

If you want to run the code yourself, please create an AWS account (https://github.com/coventry/tfintro/blob/master/notes.org#create-aws-account) ahead of time, and make sure that you can run a GPU-backed instance (https://github.com/coventry/tfintro/blob/master/notes.org#create-a-tensorflow-machine-from-scratch--ec2-instance) before the class. (Don't forget to shut it down once you've started it: cost of running it is 65c/hour.) If AWS stops you from creating it, request a limit increase (http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-resource-limits.html#request-increase), and mention this class in the request.