Machine Learning for the Properties of Exotic Nuclei

The astrophysical r-process is responsible for producing roughly half of the Universe's heaviest elements, yet its modeling is highly sensitive to nuclear masses, many of which remain experimentally inaccessible. We present a machine learning (ML) framework to predict nuclear masses across the full chart of nuclides, trained on experimental data and guided by physical principles. I will also discuss how these predicted masses influence r-process nucleosynthesis and demonstrate how observed abundance patterns can help constrain mass extrapolations in regions far from stability.