Albert Kjøller Jacobsen
DDSA PhD Fellow | Geometric Approximate Bayesian Inference
I am a DDSA PhD Fellow at the Technical University of Denmark (DTU), supervised by Georgios Arvanitidis from the Section for Cognitive Systems. My research advances approximate Bayesian inference by leveraging differential geometry. I got into this through my earlier work on geometry in optimization for improving generalization performance.
I’m particularly interested in:
- probabilistic machine learning and approximate Bayesian inference,
- deep learning theory and optimization,
- topics from geometry in machine learning.
news
| Aug 30, 2026 | I’m attending the Machine Learning Summer School at the Max Planck Institute for Intelligent Systems in Tübingen for the next two weeks. |
|---|---|
| Jul 30, 2026 | I got an honourable mention for the best poster at EDS in Lisbon. Thank you so much to everyone who stopped by! |
| Jul 26, 2026 | I’ll be attending the ELLIS Doctoral Symposium (EDS) on Trustworthy AI in Lisbon next week. Looking forward to meeting fellow PhD students as well as giving a talk and a poster presentation on my project in geometric approximate Bayesian inference. |
| Jul 04, 2026 | New preprint: “Beyond Laplace: Closed-form wrapped Gaussian posterior approximations on statistical manifolds”. Amazing work by Marcelo and I’m happy to have helped out. |
| May 18, 2026 | New preprint: “Don’t Stop Me Yet: Sampling Loss Minima via Dissipative Riemannian Mechanics”. Super excited about this work – if you like geometry, mechanics or uncertainty quantification, this one is for you. |