/For the optimization of media for protein production with yeast, we similarly observed a large distribution of values of specific productivity (Fig. 5A) with widespread spacing of experiments covering the design space in Iteration 1 and 2 (Fig. 5B). Unsurprisingly, the algorithm planned experiments in Iteration 1 for types of carbon sources not probed in the initial experiments since these regions of the design space still had the highest uncertainty, dominating the objective function. The subsequent iterations (3-6), however, emphasized exploitation as observed by the higher fraction of experiments planned in a limited region of the design space resulting in increased specific productivity (Fig. 5A) with limited testing of other regions of the design space (Fig. 5B).