Developing Interpretable Machine Learning Approaches for Understanding and Optimizing Properties of MetaMaterials
Our goal is to develop ML models that are interpretable to facilitate metamaterial design by optimizing the properties of interest. In the case of metamaterials, the design space is too large to run simulations or perform experiments for every possible design. Since the reasoning processes of interpretable ML models can be understood by a human expert, they can help identify the underlying structure-property relationship of the metamaterials. This project will focus on both static and dynamic mechanical properties with the aim to gain a better understanding of the current data, unveil the structure-property relationship, and expedite targeted material design. The student will use and develop interpretable machine learning tools and apply them to material datasets.