Clouds play a significant role in atmospheric processes and considerably affect the Earth’s radiative budget, yet they pose significant uncertainties in weather and climate models. Detailed information about cloud microphysical processes, describing the formation and interaction of individual cloud and precipitation particles, is indispensable for further improvement, yet not feasible to obtain from conventional modeling. Our project aims to develop a method based on Machine Learning to estimate micophysical process rates from standard output variables, guided by the underlying laws of physics. For this purpose, we run high-resolution simulations with the Icosahedral Nonhydrostatic (ICON) model for various meteorological situations in order to generate training and validation datasets.