Computational quantum chemistry is a useful tool to assess the properties of novel compounds, materials, and reactions. While these investigations are often valuable for the specific systems at hand, their broader impact tends to be more limited. Our work focuses on merging computational chemistry and molecular modeling with Big Data ideas. The virtual high-throughput screening of compound libraries allows us to generate huge quantum chemical data sets, which can be employed for data-driven discovery. To extract an understanding of the underlying structure-property relationships from these data sets we develop chemical data mining tools. These have their roots in machine learning, statistical learning, and informatics. We will discuss the utility of these tools and the resulting models for the prediction of molecular properties without the need for expensive quantum chemical calculations.
Back to Machine Learning for Many-Particle Systems