Generative Adversarial Networks

Datum konání: 07.03.2017
Přednášející: Jan Kotera
Odpovědná osoba: Kotera

As a followup to Michal Sorel's recent talk about generative neural networks, I will give a short and very non-exhaustive/non-technical intro to the recently very popular setup of neural networks for drawing data from distributions given by samples (read: generate "random" images that look "real" without having to specify what "real" means) with apparently great potential in unsupervised learning and related applications.