Chemical Reaction Optimization: Multi-Task Learning

This lecture by Kobi Felton explores the use of multi-task learning to accelerate chemical reaction optimization. The presentation covers the challenges of scaling up reactions, the vast parameter space in chemistry, and the limitations of experimental resources. Felton introduces automated optimization workflows and optimization algorithms, showcasing examples of N-benzylation and C-N cross-coupling reactions. The lecture emphasizes the importance of global optimization algorithms in chemistry and demonstrates the application of Bayesian optimization in reaction optimization. Various strategies and benchmarks are discussed, highlighting the benefits and potential failures of multitask Bayesian optimization in reaction optimization.