Machine learning and experiments: A synergy for the development of functional materials

With machine learning (ML) and artificial intelligence (AI) becoming increasingly refined and accessible, computer engineers and materials scientists are utilizing these data-driven techniques to design new functional materials more efficiently. Additionally, the advancement of simulation software and computing power has substantially lowered the cost of obtaining training data. However, using only simulation data presents a difficulty in the eventual realization of a material design due to possible misalignment of the simulation setup and physical laboratory conditions. Therefore, it is mutually beneficial to also improve the experimental aspect of functional materials development using ML and AI techniques. In this article, we survey the current state of ML/AI involvement in functional materials design, focusing specifically on acoustic/mechanical metamaterials, piezoelectric materials, and biological materials. The macroscopic nature of these functional materials lends well to additive manufacturing fabrication, which makes optimizing the synthesis process of these materials highly desirable. We conclude by pointing out a few promising directions for future investigation of functional materials and their place in societal applications. Graphical abstract