Machine Learning with Enormous "Synthetic" Data Sets: Predicting Glass Transition Temperature of Polyimides Using Graph Convolutional Neural Networks.

In the present work, we address the problem of utilizing machine learning (ML) methods to predict the thermal properties of polymers by establishing "structure-property" relationships. Having focused on a particular class of heterocyclic polymers, namely polyimides (PIs), we developed a graph convolutional neural network (GCNN), being one of the most promising tools for working with big data, to predict the PI glass transition temperature as an example of the fundamental property of polymers. To train the GCNN, we propose an original methodology based on using a "transfer learning" approach with an enormous "synthetic" data set for pretraining and a small experimental data set for its fine-tuning.

Transport Properties of Thermoplastic R-BAPB Polyimide: Molecular Dynamics Simulations and Experiment.

The present work evaluates the transport properties of thermoplastic R-BAPB polyimide based on 1,3-bis(3,3',4,4'-dicarboxyphenoxy)benzene (dianhydride R) and 4,4'-bis(4-aminophenoxy)biphenyl (diamine BAPB). Both experimental studies and molecular dynamics simulations were applied to estimate the diffusion coefficients and solubilities of various gases, such as helium (He), oxygen (O), nitrogen (N), and methane (CH). The validity of the results obtained was confirmed by studying the correlation of the experimental solubilities and diffusion coefficients of He, O, and N in R-BAPB, with their critical temperatures and the effective sizes of the gas molecules, respectively.