Fig. 1: Schematic representation of ANN training used in this work.
From: Towards AI-driven autonomous growth of 2D materials based on a graphene case study

After initialization with the parameter guess, the protocol generation is conducted (i), followed by sample growth (ii). The obtained sample is then characterized through Raman spectroscopy (iii). The extracted data are used to evaluate the score (iv). Finally, the protocol is updated with the new parameters (v) and a new protocol is generated.