================================================================================

Machine-guided optimization to apply artificial intelligence to chemistry

Hyun Woo Kim
Chemical Data-Driven Research Center, Korea Research Institute of Chemical Technology (KRICT), Daejeon

34114, Korea

Artificial intelligence (AI), including machine learning (ML), is gaining a lot of attention in recent years as
more and more data become available in chemistry-related fields. Moreover, it demonstrates its ability to find
hidden relationships in the data. For example, physical properties of molecules are reasonably predicted with
several ML techniques, and well-established prediction models can be used to suggest promising candidates for
new molecules and materials. Here, I will present our recent work to optimize molecular representations and
reaction conditions in a data-driven way. My talk will be started by introducing the basics of data-driven
optimization. We will then see how to process the data, along with some description of the ML code. By doing
so, we will understand the need for machine-guided representations to improve the performance of ML in
predicting molecular properties. After that, I will show an application of AI to suggest better reaction conditions
of a methane conversion reaction. Finally, I will briefly mention a brand new platform for using ML built by
our institute.