Tolles Video [5min43], muss man gesehen haben. Das Video beschreibt eine neue KI-Methode des Deep Learnings:
Ausgehend von einer niedrigen Auflösung fügt das System neue Ebenen hinzu, die im Verlauf des Trainings immer feinere Details modellieren.
Das Ergebnis sind «Photos» von Menschen, welche es gar nicht gibt.
Die Forschung bei NVIDIA treibt Deep Learning voran:
Generating photos of wounderful people who do not exist. Great video, must see.
Abstract: 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.
Authors:
Tero Karras (NVIDIA), Timo Aila (NVIDIA), Samuli Laine (NVIDIA), Jaakko Lehtinen (NVIDIA and Aalto University).
Source : Youtube
More Info about NVIDIA deep learning applications and results:
https://blogs.nvidia.com/