Generative Adversarial Networks (GANs) are a recent approach to generating realistic images, natural language, and other data. They were conceived by Ian Goodfellow in 2014 and have exploded in popularity in the deep learning community.
The most impressive example of GANs has been Progressive GANs by NVIDIA - in their video they smoothly generate artificial examples of a diverse selection of celebrity faces in exceptional detail.
Learn how to build and train your own GANs at our session.
GANs are surprisingly easy to build, but difficult to train. We will also learn about some of the challenges with training GANs and some of the tricks deep learning researchers have used to succeed in their training.
Researchers have been working on many exciting ideas for improving GANs, and we will also learn about some of the more interesting ideas.