Machine learning has significantly improved the way cosmologists model and interpret large-scale structure data. In this talk, I will introduce an explainable deep learning framework designed to capture the complexities of non-linear large-scale structure. This model relies on a minimal set of physically interpretable parameters, and generalizes to phenomena beyond its specific training setting. I will focus on applications related to dark matter halos, which form the building blocks of the large-scale structure and wherein galaxy formation takes place. The goal is to use interpretable neural networks to model final emergent properties of dark matter halos, such as their density profiles, and connect them to the physics that determines those properties.
