The multiscale design of soft materials requires an ensemble of computational techniques spanning quantum-chemistry to molecular dynamics to continuum modeling. The recent emergence of machine-learning (ML) and modern optimization algorithms has accelerated material property prediction, as well as stimulated the development of hybrid ML/molecular modeling methodologies capable of providing physical insights unobtainable from purely physics-based modeling and intuition. Such hybrid techniques also have important ramifications for the ML-enhanced interpretation of results from simulations and experiments alike. Leveraging ML techniques for the design of chemical or morphological structures based on a target property or functionality represents an exciting goal for the general area of soft materials, including polymers, liquid crystals, colloids, or biomolecules, to name a few representative classes of systems. Here, we provide a perspective on recent work using ML techniques of relevance for the multiscale design of soft materials and outline potential future directions of interest to the soft materials community.

软材料的多尺度设计需要一系列的计算技术，从量子化学到分子动力学再到连续模型。机器学习（ML）和现代优化算法的最新出现加速了材料特性的预测，并刺激了混合ML /分子建模方法的发展，该方法能够提供纯粹基于物理的建模和直觉无法获得的物理见解。这样的混合技术对于从仿真和实验中得到的ML增强的结果解释也具有重要的意义。利用ML技术基于目标特性或功能来设计化学或形态结构，代表了软材料（包括聚合物，液晶，胶体，或生物分子，仅举几例代表性的系统。在这里，我们对使用与机器学习材料相关的ML技术的多尺度设计的最新工作提供了一个观点，并概述了对软材料领域感兴趣的潜在的未来方向。