Our team is currently developing two projects focused on applying machine learning to predict the properties of chemical compounds.
RGBChem – A method that translates chemical information into RGB images later used to train convolutional neural networks. GitHub Article
PICle – A procedure that, through the use of machine learning, makes it possible to predict the potential energy of bimetallic clusters and, as a result, analyze catalytic pathways without the need for expensive calculations. GitHub Article