Extraction of organic chemistry grammar from unsupervised learning of chemical reactions

By CASC4DE April 16, 2021

 

A significant step forward to automate atom mapping without human supervision using Transformer Neural Network and training set of reaction SMILES.

Philippe Schwaller and co-workers recently published an outstanding approach.

Authos succeed in extracting coherent chemical grammar from unannotated sets of reactions.

Read more on science.org