This Website is not fully compatible with Internet Explorer.
For a more complete and secure browsing experience please consider using Microsoft Edge, Firefox, or Chrome

Generative AI for Simulation of Fluids and Inverse Design

Generative learning has been receiving much attention recently, mostly due to the capabilities in image and text generation. The basis of generative learning is data that follows a certain statistical distribution. Using this data, generative learning is trained on the task to sample from the statistical distribution and thus produce new samples that never were part of the training data set, but share all the statistically relevant features. The important point with generative learning is scalability: with sufficient data at hand, generative learning learns distributions as complex as faces of persons or even all sorts of arbitrary images of different types on the internet.

Especially given the last very general statistical distribution, narrowing in to certain sub-domains of images via (text based) conditioning is crucial. Conditional generative learning thereby is key to the actual applicability of the technology.

This article appeared in the January 2024 issue of BENCHMARK.

Document Details

AuthorsDrygala. C Gottschalk. H Kruger. P di Mare. F Krebs. W Werdelmann. B
TypeMagazine Article
Date 10th January 2024
OrganisationsTechnical University Berlin Ruhr University Bochum Siemens


Back to Previous Page