Machine learning can accelerate the screening, design and discovery of new molecules and materials in multiple ways, e.g. by virtually predicting properties of molecules and materials, by extracting hidden relations from large amounts of simulated or experimental data, or even by interfacing machine learning algorithms for autonomous decision-making directly with automated high-throughput experiments. In this talk, I will focus on our research activities on graph neural networks for property prediction [1] and inverse design of crystal structures using Bayesian flow networks [2], as well as on the use of active learning with Bayesian optimization for automated data analysis and autonomous decision-making in self-driving labs, especially in the area of organic semiconductors for photovoltaics [3,4]. 

[1] Reiser et al., Communications Materials 3, 93 (2022), https://www.nature.com/articles/s43246-022-00315-6
[2] Under review in ICML 2025
[3] Wu et al., JACS 145, 30, 16517–16525 (2023), https://pubs.acs.org/doi/full/10.1021/jacs.3c03271
[4] Wu et al., Science 386, 6727, 1256-1264024 (2024), https://www.science.org/doi/abs/10.1126/science.ads0901