Skip to main content
Fig. 11 | Journal of Cheminformatics

Fig. 11

From: CIME4R: Exploring iterative, AI-guided chemical reaction optimization campaigns in their parameter space

Fig. 11

LineUp overview of the reactions performed, their measured yields, and calculated SHAP values, grouped by cycle (0–7) and ordered by measured yield. After a random initialization in cycle 0, EDBO prioritized experiments with the currently best ligand (A) for the next three cycles. In cycle 4, EDBO explored different ligands, but for fixed base and solvent (C). Finally, EDBO managed to find a yield of 100% by exploiting the best-performing yield found in cycle 4 (D). EDBO’s choice of parameters and the development of the SHAP values (B15, E) aid users in understanding the model’s decisions (T4)

Back to article page