Anton Baumann

anton@antonbaumann.com - scholar - github - linkedin

machine learning researcher working on probabilistic methods, vision-language models and sample efficient reinforcement learning

Selected publications

Reinforcement Learning via Self-Distillation
- Jonas Hübotter, Frederike Lübeck^*, Lejs Behric^*, Anton Baumann^*, Marco Bagatella, Daniel Marta, Ido Hakimi, Idan Shenfeld, Thomas Kleine Buening, Carlos Guestrin, Andreas Krause
- ICML 2026; ^*equal second authorship.
- Best Paper Award at ICLR 2026 Workshop on Test-Time Updates and Oral Presentation at ICLR 2026 Workshop on Scaling Post-training for LLMs.
- Preprint - Project - GitHub
Post-hoc Probabilistic Vision-Language Models
- Anton Baumann, Rui Li, Marcus Klasson, Santeri Mentu, Shyamgopal Karthik, Zeynep Akata, Arno Solin, Martin Trapp
- ICLR 2026
- Preprint - Project - GitHub
Probabilistic Active Few-Shot Learning in Vision-Language Models
- Anton Baumann, Marcus Klasson, Rui Li, Arno Solin, Martin Trapp
- NeurIPS 2024 BDU (spotlight & poster)
Probabilistic MIMO U-Net
- Anton Baumann, Thomas Roßberg, Michael Schmitt.
- ICCV 2023 UnCV (oral & poster)
- Preprint - GitHub

Research roles and applied ML work across academia and industry.

Aalto University, Helsinki
- Summer Research Intern, Machine Learning Research Group (May 2024 - May 2025)
  - Researched probabilistic ML for vision-language models in Arno Solin's lab.
Bundeswehr University, Munich
- Research Assistant, Earth Observation Lab (Feb 2023 - May 2024)
  - Applied probabilistic ML and data fusion to remote sensing problems.
molab.ai, Munich
- Data Scientist (Internship, then Part-time) (Aug 2022 - Feb 2023)
  - Implemented GNNs for chemical property prediction in PyTorch.
Ablacon GmbH, Munich
- Researcher (Working Student), then Bachelor's Thesis (Mar 2020 - May 2022)
  - Extended Horn & Schunck optical flow for electrographic flow mapping.
  - Integrated the method into a prototype core product with a research engineering team.
Cliqz GmbH, Munich
- Full-stack Developer (Working Student) (Sep 2017 - Jul 2019)
  - Developed logging/versioning system and JSON diff library in Go for privacy-first ad-tech infrastructure.

ETH Zürich
- Master's Thesis in Computer Science (Oct 2025 - Jun 2026 expected)
  - Meta test-time training for LLMs with joint supervision from Andreas Krause (ETH Zürich) and Zeynep Akata (TUM).
  - Day-to-day supervision by Jonas Hübotter.
KTH Stockholm
- Exchange Semester (May 2024 - Feb 2025)
Technical University of Munich
- M.Sc. Computer Science (Oct 2023 - Jun 2026 expected)
  - Ranked 1st/202 in official semester ranking (current GPA 1.2).
- B.Sc. Computer Science, Minor in Mathematics (Oct 2018 - May 2022)
  - Thesis: "EGFX: Determining Optical Flow on Arbitrary Surfaces." (supervised by Daniel Cremers; GPA 1.0)
- Early-study program in Computer Science (Oct 2016 - Oct 2017)