AI and Catalysis

Catalysts are the backbone of industrial chemistry. They speed up reactions and make possible what was once impossible. Developing new catalysts is hard because as well as the reagents in the primary reaction there are often intermediate reactions and co-reagents. Understanding and optimizing what is occurring is a highly multi-dimensional challenge. This paper explains and gives examples of how AI can be used to efficiently explore these dimensions and how hierarchical, graphical AI models can be used to model complex interactions and compensate for scarce data through transfer learning.