Data and Machine Learning in Polymer Science

Abstract

Data-driven innovation has shown great power in solving problems in multifactor correlation, convergence and optimization, synergistic and antagonistic effects, pattern and boundary identification, critical behavior and phase transition, which are ubiquitous in polymer science. Either for the in-depth understanding of physical problems or in the discovery of new polymer materials, integrating data and machine learning into conventional experimental, theoritical, modeling and simulation approaches becomes blooming. Here we present a perspective based on our research interests, highlight some key issues and provide a prospection in this emerging direction. We focus on a number of typical advances in the description and identification of polymer conformation and structures, and the interpretation and prediction of structure-property correlations, that have applied data and machine learning in polymer science.