Lee Nicolas A, Shen Sabrina C, Buehler Markus J
Laboratory for Atomistic and Molecular Mechanics (LAMM), Massachusetts Institute of Technology, 77 Massachusetts Avenue 1-165, Cambridge, MA 02139, USA.
School of Architecture and Planning, Media Lab, Massachusetts Institute of Technology, 75 Amherst Street, Cambridge, MA 02139, USA.
Matter. 2022 Nov 2;5(11):3597-3613. doi: 10.1016/j.matt.2022.10.003.
Recently, the potential to create functional materials from various forms of organic matter has received increased interest due to its potential to address environmental concerns. However, the process of creating novel materials from biomass requires extensive experimentation. A promising means of predicting the properties of such materials would be the use of machine-learning models trained on or integrated into self-learned experimental data and methods. We outline an automated system for the discovery and characterization of novel, sustainable, and functional materials from input biomass. Artificial intelligence provides the capacity to examine experimental data, draw connections between composite composition and behavior, and design future experiments to expand the system's understanding of the studied materials. Extensions to the system are described that could further accelerate the discovery of sustainable composites, including the use of interpretable machine-learning methods to expand the insights gleaned from to human-readable materiomic insights about material process-structure-functional relationships.
最近，利用各种形式的有机物质制造功能材料的潜力因其解决环境问题的潜力而受到越来越多的关注。然而，从生物质中制造新型材料的过程需要大量实验。一种预测此类材料性能的有前景的方法是使用基于自学实验数据和方法训练或集成的机器学习模型。我们概述了一个自动化系统，用于从输入的生物质中发现和表征新型、可持续和功能性材料。人工智能提供了检查实验数据、建立复合材料组成与性能之间联系以及设计未来实验以扩展系统对所研究材料理解的能力。文中描述了该系统的扩展，这些扩展可以进一步加速可持续复合材料的发现，包括使用可解释的机器学习方法将从材料过程 - 结构 - 功能关系中获得的见解扩展为人类可读的材料组学见解。
