Inhalt des Dokuments
A short intro
In 2019 TU-Berlin and the chemical company BASF SE founded BASLEARN, the Berlin based Joint Lab for Machine Learning.
Embedded on the Machine Learning / Intelligent Data Analysis group led by Prof. Klaus Robert Müller, BASLEARN mission is to foster collaboration on research for challenges derived from chemistry and chemical engineering field by developing novel Machine Learning algorithms.
With offices located at the MAR building at Marchstr. 23 and at BasCat building at Hardenbergstr. 36, we want to bring closer academia and industry, contributing with novel ideas and project extracting the full potential of such colaboration.
The exploration of chemical compound space is crucial to discover
new molecules and materials for certain purposes, e.g. in catalysis,
in drug design, or for generation and storage of renewable energy.
Machine learning (ML) is perfectly suited to accelerate this process, which requires extensive simulations and tedious evaluation of chemical properties.
Our group focuses on the development of ML algorithms for quantum-mechanical, atomistic systems.
We build domain specific models which incorporate fundamental physical constraints and prior-knowledge to tackle various sorts of problems.
This includes deep neural networks as well as kernel Hilbert space methods to predict quantum-chemical properties and enable fast molecular dynamics simulations at chemical accuracy.
Furthermore, we investigate generative models which learn to sample unseen molecules or materials with desired properties given a set of example structures.