- Mar 27, 2017
That's rather creepy actually, but kind of cool at the same time.MSN has just published an article on the experience created by Philip Wang, engineer at Uber (The company now participates in the training of artificial AI). His site uses StyleGAN, a tool created by Nvidia.Generative adversarial networks, a class of algorithms that work by machine learning. "In this case, a first program tries to create face images from raw data, then a second program appreciates these images and decides whether they are realistic or not. If the result is satisfactory, it is informed that it is on the right track.This game of ping-pong between the two programs improves the result, and get, in fine, credible photos."
If you go to the site it will show only one photo at a time and you have to reopen the site to display a new face: