Machine learning-driven new material discovery

The combination of physicochemical laws and empirical trial and error has long guided the design material. The space of hypothetical materials to be considered is incredibly large, and only a small fraction of possible compounds can ever be tested experimentally. The computational techniques of atomistic simulation and machine learning (ML) offer an avenue to rapidly invent new materials and navigate this enormous space. The recently proposed crystal graph convolutional neural network (CGCNN) offers a highly versatile and accurate machine learning (ML) framework by learning material properties directly from crystal graph representations. In this talk, we discuss the identification of high-quality candidates by CGCNN ML about physicochemical properties of delafossite and perovskite materials in various element combinations. For distinct material- and application-specific machine learning, we generated appropriate descriptors (the sets of parameters capturing the underlying mechanisms of a material's property) by benchmarking the sure independence screening and sparsifying operator (SISSO).