Bergische Universität Wuppertal
Fakultät für Mathematik und Naturwissenschaften
Wissenschaftliches Rechnen und Hochleistungsrechnen
42119 Wuppertal, Germany
Phone: +49 202 439 5192
Fax: +49 202 439 3135
Email: holzenkamp at uni-wuppertal.de
Office: Building MI, Room: (tba)
Matthias Holzenkamp is a Ph.D. student, who works in the area of molecular machine learning. After a bachelor’s degree in mechanical engineering and a master’s degree in physical engineering at TU Berlin, he has a broad background in computer simulations.
His research is part of the DFG Priority Program SPP 2363 which has the goal to develop and apply machine learning algorithms to molecular problems. The project focuses on active learning approaches for properties of molecule geometries as well as geometries of molecular systems: Since these properties are often needed for a large number of geometries, and the calculation with ab initio methods can be very expensive, it is straightforward to replace these calculations with ML models that map geometries to the desired properties. The goal now is to keep the number of required ab inito calculations as small as possible by actively finding the most suitable training data.
Find the project as well as all other SPP projects under: https://www.uni-muenster.de/SPP2363/Projects/projects.html
Another research interest is tensor networks and their applications in machine learning. In his master thesis, he used the tensor train format to perform model selection and prevent the curse of dimensionality in high-dimensional feature spaces.