Fig. 2: Flow-diagram for the proposed AI-driven end-to-end optimization framework.

The framework inputs include constant column vector (i.e., q vector) and a nodal point property matrix (i.e., P matrix). The framework is composed of three data sampling algorithms, whose objective is to generate instances of microstructure representations, i.e., the multidimensional ODFs.