Nils Candebat   (France)

nils.candebat @ uni-heidelberg.de

Study of stellar cluster in nearby galaxies through machine learning

Hi, I'm Nils Candebat, a PhD student at Institut für Theoretische Astrophysik der Universität Heidelberg. My research focuses on the intersection of machine learning and astrophysics. My doctoral thesis I conduct as a member of the Star Formation Group at ITA and the international PHANGS collaboration.

While AI has been widely adopted across many fields, the astrophysics community has not yet fully embraced these. Several factors contribute to this hesitation:

 *      The domain gap that makes AI less efficient across different astronomical instruments
 *      Challenges in uncertainty quantification
 *      Adaptation difficulties
 *      Limited access to GPU resources

My goal is to bridge this gap and accelerate the adoption of AI in astrophysics. I primarily work with conditional Invertible Neural Networks, which are Bayesian Networks that enable robust uncertainty quantification.  Additionally, I'm exploring how Large Language Models (LLMs) can be leveraged to create foundation models that will be used as the backbone of future astronomical research.

Supervisor:    Ralf Klessen   (ITA)