Previous talks at the SCCS Colloquium

Josef Mayr: Bayesian Optimization of Material Synthesis Parameters with Gaussian Processes

SCCS Colloquium |


Bayesian optimization has been found to be a valuable tool for efficient global optimization of expensive black-box functions. Such expensive functions can be found, for example, in chemistry where a reaction is to be performed in the real world. A desirable goal is to minimize number of trials required to find an experiment that satisfies certain criteria.

The goal of this work is to apply Bayesian optimization to reaction optimization in the context of metal-organic frameworks to find suitable reaction conditions. In addition, the approach is augmented by an approximation of the reaction process via the solution of a differential equation describing the reaction process.

Guided research presentation. Josef is advised by Dr. Felix Dietrich.