AI-Paper von Tero Karras mit einer neuen Methode für das Training von Neural Networks, mit der sie HighRes-Bilder ausrechnen können, in dem Fall einen Celebrity-Facespace.

Zur Erinnerung: Das sah vor ein paar Wochen noch so aus.

We describe a new training methodology for generative adversarial networks. The key idea is to grow both the generator and discriminator progressively: starting from a low resolution, we add new layers that model increasingly fine details as training progresses. This both speeds the training up and greatly stabilizes it, allowing us to produce images of unprecedented quality, e.g., CELEBA images at 10242.

Facespace auf Nerdcore:
20 Minutes thru Face Space
Visual AI-Spaces Auto-Pilot
Human Fetuses prefer „face-like“ Shapes
Algo-Faces reconstruced from Monkey Brainwaves
Neural Facegrid 256


Wouldn't fly:

Wouldn't pet:

Definitely wouldn't pet:

Come on…