“…As briefly anticipated in the introduction, ML can be a powerful tool for material science and design, owing to its ability to efficiently find patterns and trends, handle multidimensional and heterogeneous data and improve continuously with access to more observations-all while being intrinsically and fully automatable. Models and simulations based on ML approaches can, for instance, quickly narrow down the parameter space of specific variables involved in fabrication processes [18,181,182]. They can also be used to control, tweak or even design ad hoc properties of materials [131], heterostructures [9,132] and devices [19, 136,137], again while being suitable for fabrication strategies that require large-scale, fast and automated production.…”