Generative models with neural network approximations have shown impressive
results in many applications, in particular in inverse problems and data assimilation.
First developments in this direction were
generative adversarial networks by Goodfellow et al. in 2014
and variational autoencoders by Kingma and Welling in 2013.
Recently, flow based models as normalizing flows, score based diffusion models and flow matching
have attained great interest and provide state-of-the art results.
In the seminar we will mainly focus on flow matching techniques,
but will start with a dynamic view onto optimal transport which is closely related to the topic.
- Trainer/in: Gabriele Steidl