Machine Learning in Chemistry: Bayesian Reaction Optimization

This lecture explores the application of machine learning in chemistry, focusing on Bayesian reaction optimization using a robotic flow synthesis platform. The instructor discusses shifting the experimental burden from humans to machines, computer-aided synthesis planning, continuous flow chemistry, and the DARPA Make-It program. The lecture covers the integration of analytics, the closed-loop experimentation cycle, and the optimization of multistep routes and downstream processes. Various analytical techniques, automation gaps, and the use of Bayesian optimization algorithms are highlighted. Real-time FT-IR data, multi-objective optimization, and the visualization of response surfaces are also discussed, along with the project summary and future outlook.