A prospective on machine learning challenges, progress, and potential in polymer science

作者：Daniel Struble, Bradley G. Lamb, Boran Ma · 发表于：MRS Communications · 年份：2024 · DOI：10.1557/s43579-024-00587-8 · 被引用次数：29 · 研究领域：Machine Learning in Materials Science、Computational Drug Discovery Methods、Scientific Computing and Data Management

Abstract Artificial intelligence and machine learning (ML) continue to see increasing interest in science and engineering every year. Polymer science is no different, though implementation of data-driven algorithms in this subfield has unique challenges barring widespread application of these techniques to the study of polymer systems. In this Prospective, we discuss several critical challenges to implementation of ML in polymer science, including polymer structure and representation, high-throughput techniques and limitations, and limited data availability. Promising studies targeting resolution of these issues are explored, and contemporary research demonstrating the potential of ML in polymer science despite existing obstacles are discussed. Finally, we present an outlook for ML in polymer science moving forward. Graphical Abstract