Premium
Prediction of Polymer Properties from their Structure by Recursive Neural Networks
Author(s) -
Duce Celia,
Micheli Alessio,
Starita Antonina,
Tiné Maria Rosaria,
Solaro Roberto
Publication year - 2006
Publication title -
macromolecular rapid communications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.348
H-Index - 154
eISSN - 1521-3927
pISSN - 1022-1336
DOI - 10.1002/marc.200600026
Subject(s) - representation (politics) , polymer , macromolecule , encoding (memory) , methyl methacrylate , artificial neural network , methacrylate , tree (set theory) , polymer science , tree structure , biological system , process (computing) , computer science , materials science , polymer chemistry , algorithm , copolymer , chemistry , artificial intelligence , mathematics , composite material , combinatorics , biochemistry , politics , political science , binary tree , law , biology , operating system
Summary: We propose a new approach for predicting polymer properties from structured molecular representations based on recursive neural networks. To this aim, a structured representation is designed for the modeling of polymer structures. This representation can also account for average macromolecule characteristics. Preliminarily, this model is applied to the calculation of the T g of (meth)acrylic polymers with different stereoregularity.Representation of poly(methyl methacrylate) as a chemical tree and unfolding of the encoding process through its structure.