Deep learning techniques teach neural model to 'play' retrosynthesis

Chemical engineers have trained a neural network model to plan synthetic routes to any target molecule, optimizing user-specified objectives such as cost, time, and sustainability.