Recursive Neural Networks for Quantitative Structure-Property Relationship Analysis of Polymers

摘要

We report an overview of the results obtained in the prediction of the glass transition temperature of a set of acyclic acrylic and methacrylic polymers by Recursive Neural Networks (RecNN). The RecNN approach allows us a completely novel approach to QSPR/QSAR analysis by a direct adaptive processing of molecular graphs, both in case of small molecules or polymers.