Reinforcement Learning

Reinforcement learning is a widely used approach in ML that rewards a model for desirable perfomance (e.g. accurate prediction of chemical reactions) and penalizes it for undesired ones. Researchers in C-CAS and their external partners use this approach to find optimal coditions with general applicability, e.g. in C-H arylation reactions.

Publications

• Wang, J. Y.; Stevens, J. M.; Kariofillis, S. K.; Tom, M.-J.; Li, J.; Tabora, J. E.; Parasram, M.; Shields, B.; Primer, D.; Hao, B.; Del Valle, D.; DiSomma, S.; Furman, A.; Zipp, G. G.; Melnikov, S.; Paulson, J.; Doyle, A. Reinforcement learning prioritizes general applicability in reaction optimization.  ChemRxiv 2023 10.26434/chemrxiv-2023-dcg9d