Machine learning comprises a set of techniques by which structure and correlations are extracted from data in an automated fashion. The derived models are then used to make predictions about new situations not seen in the data. In the past decade, machine learning has turned from a niche discipline to a technology that is influencing more and more parts of our daily lives and the way we do science.
In computational chemistry, a particularly successful application is the development of machine learning potentials, which vastly extend the time and length scales accessible to molecular simulations while mostly preserving the accuracy of the underlying more expensive methods (like DFT or CC) that were used to generate the data.
We are in particular interested in graph neural networks (GNNs), which are the state of the art in quantum-chemical machine learning. Our initial efforts were a contribution to using natural bond orbital (NBO) and similar information from quantum chemical calculations as inputs for those neural networks. Furthermore, we successfully used k-medoids clustering (a type of unsupervised machine learning) to extract molecular structure information from non-Born–Oppenheimer wavefunctions.
This research theme started in connection with Lucas’ Marie Skłodowska-Curie postdoctoral fellowship funded by the European Union (“ML4Catalysis”, grant number 101025672).
Related publications:
- Hannes Kneiding, Ruslan Lukin, Lucas Lang, Simen Reine, Thomas Bondo Pedersen, Riccardo de Bin and David Balcells, Deep learning metal complex properties with natural quantum graphs, Digital Discovery 2, 618 (2023).
- Lucas Lang, Henrique Musseli Cezar, Ludwik Adamowicz and Thomas Bondo Pedersen, Quantum Definition of Molecular Structure, J. Am. Chem. Soc. 146, 1760 (2024).