Molecular Representation: Going Long on Fingerprints
Author(s)Pattanaik, Lagnajit; Coley, Connor Wilson
MetadataShow full item record
Machine learning for chemistry requires a strategy for representing (featurizing) molecules. In this issue of Chem, Sandfort et al. describe an approach that concatenates 24 fingerprint representations into 71,375-dimensional vectors, which are then used for a variety of supervised learning tasks related to chemical reactivity.
DepartmentMassachusetts Institute of Technology. Department of Chemical Engineering
Pattanaik, Lagnajit and Connor W. Coley. "Molecular Representation: Going Long on Fingerprints." Chem 6, 6 (June 2020): 1204-1207. © 2020 Elsevier Inc
Author's final manuscript