ML for molecular applications

The development of functional molecules (e.g. pharmaceuticals, agrochemicals) is a central task of chemical research. Due to the complexity of the underlying relationships, data-driven modeling using ML is an indispensable tool. Here, students are given the opportunity to acquire AI competences that enable them to critically discuss ML methodologies as well as current related AI developments. For this purpose, the basics of molecular ML (chemoinformatics, representation of chemical structures, model development and validation) will be taught first. Since availability and quality of source data are crucial, aspects of data science will also be addressed. Subsequently, selected chemical ML applications will be discussed, e.g. methods for the prediction of molecular properties and bioactivity, as well as ML methods for synthesis planning or reaction optimization. A core component is the practical application of the learning content, e.g. using Jupyter notebooks. The aim is to provide students with tools that enable them to independently evaluate AI methodologies in the context of their own research.