Talks
-
SympFormer: Accelerated attention blocks via Inertial Dynamics on Density Manifolds
- Oberwolfach workshop Flows on Measure Spaces and Applications in Machine Learning.
- Accelerated Stein Variational Flow
- GSI’25 – 7th International Conference on Geometric Science of Information, St. Malo, France, 30.10.2025. Slides
- Stan Osher’s UCLA Level Set Seminar, April 2025.
- Colloquium of the Signal Processing Department, University Carlos III Madrid, November 2025.
- Wasserstein gradient flows of MMD-regularized f-divergences
- Conference on Mathematics of Machine Learning 2025, Hamburg, 30.09.2025. Slides · Video
- Stan Osher’s UCLA level set seminar, 19.08.2024. Slides · Video
- Poster at Workshop on Optimal Transport - from Theory to Applications. Pitch · Poster
- Poster at the conference Learning and Optimization in Luminy 2024.
- Poster at the SIGMA workshop 2024.
- Interpolating between Optimal Transport and KL regularized Optimal Transport using Rényi Divergences
- Wasserstein Gradient Flows of MMD Functionals with Distance Kernel and Cauchy Problems on Quantile Functions
- Joint Applied and Computational Mathematics and RTG Data Science Seminar, University of South Carolina, 30.08.2024. Slides
Expository talks
-
Training Neural Networks at the Edge of Stability
Slides
An overview of Self-Stabilization: The Implicit Bias of Gradient Descent at the Edge of Stability and Gradient Descent on Neural Networks Typically Occurs at the Edge of Stability held in Benjamin Gess’ Mathematics of Machine Learning seminar. -
What does Adam have to do with symplectic manifolds? The geometry behind momentum methods in machine learning
14th BMS Student Conference
Slides -
What are … generative probability flows for sampling?
in the “What is … ?” series of the Berlin Mathematical School
Slides -
Mode collapse and metastability in transformers
Slides
An overview of The emergence of clusters in self-attention dynamics and A Unified Perspective on the Dynamics of Deep Transformers held in Benjamin Gess’ Mathematics of Machine Learning seminar, Berlin-Leipzig Hybrid Seminar SS25. -
“Zoom and enhance” geht wirklich – mit Atomic Norm Minimization
Talk in German on my bachelor’s thesis “Atomic Norm Minimisation for Superresolution”
17. Dies Mathematicus, TU Berlin, 25.11.2022
Slides · Video · MATLAB code -
3 minute talk about the Kullback-Leibler Divergence Seminar Optimal Transport, Prof. Gabriele Steidl, summer semester 2021. Slides
-
Adversarial Regularisation in Inverse Problems
on the NeurIPS 2018 paper by Sebastian Lunz et al., seminar “Neural Networks for Inverse Problems”, winter semester 2021/2022. Slides -
Compactly supported shearlets are optimally sparse
seminar Applied Functional Analysis, summer semester 2019. Slides -
Atomic Norm Minimisation for Superresolution
seminar Optimal Transport, summer semester 2021. Slides