2020
Review on principles and algorithms and recent progress for model based Design of Experiments (DoE) to arrive at optimal experimental plans. Currently, statistical DoE is still used to plan high-throughput experimentation (HTE). The progress in model-based DoE makes it highly promising to implement this in HTE, since machine-learned models generally identify non-linear functional dependencies between the composition of a catalyst and its properties in chemical reactions. In future work, chemical expert knowledge combined with machine learning methods should be integrated in order to arrive at model approaches that are able to make reliable predictions about catalysts behaviour.
