ML-Talk am 09.07.2025: Dr. Maja Rudolph (University of Wisconsin–Madison), "Efficient Integrators for Diffusion Generative Models"
- Event Location: RPTU in Kaiserslautern, Gebäude 48, Raum 208
Event Date: 09.07.2025
Contact Person: Prof. Dr. Marius Kloft
The Machine Learning Group of Prof. Dr. Marius Kloft
at the Department of Computer Science Department at RPTU Kaiserslautern-Landau
cordially invites you to a talk on
"Efficient Integrators for Diffusion Generative Models"
by Dr. Maja Rudolph (University of Wisconsin–Madison)
on
Wednesday, July 9th 2025 at 14:30 in Room 48/208
Abstract:
Diffusion models have shown remarkable results in generative modeling, but they come with a major drawback:
slow sampling at inference time. In this talk, I’ll present two complementary strategies to accelerate sampling
in pre-trained diffusion models — Conjugate Integrators and Splitting Integrators.
Conjugate integrators generalize DDIM by mapping the sampling dynamics to a space
where they can be solved more efficiently.
Splitting integrators, inspired by techniques from molecular dynamics, reduce numerical errors
by alternating updates between data and auxiliary variables.
After exploring the theoretical foundations and empirical performance of both methods,
I’ll introduce a hybrid approach that combines their strengths.
When applied to Phase Space Langevin Diffusion on CIFAR-10, our method outperforms baselines
in both speed and sample quality, achieving high-quality samples in fewer steps.
Bio:
Starting in the Fall of 2025, Maja will be an Assistant Professor of Statistics at the University of Wisconsin–Madison,
where she works at the intersection of machine learning, probabilistic modeling, and AI.
Her research focuses on developing reliable and efficient learning algorithms,
with an emphasis on scientific and medical applications.
Maja holds a PhD in Computer Science and an MS in Electrical Engineering from Columbia University,
and a BS in Mathematics from MIT. Prior to joining academia, she served as Lead Research Scientist
at the Bosch Center for AI, where she led technical strategy on foundation models and hybrid modeling,
contributing to over 30 patent applications across industrial AI use cases.
We plan to broadcast the talk as a live stream additionally (not yet guaranteed),
but we would prefer you to attend in person:
https://sci-cast.cs.uni-kl.de/talk-stream.html
This is a scientific lecture as part of the Computer Science Colloquium, which will be held in English.
The KI-Allianz Rheinland-Pfalz supported the organization of the event.
-- Steffen Reithermann
Dr. Maja Rudolph giving her presentation