Pathway from Scientific Research

Knowledge graph for the prediction of target properties (purple) from structural features (green for interface, orange for IT-4F bulk) and from processing conditions (black), as obtained by hierarchical minimum Redundancy Maximum Relevance embedded Gaussian Process Regression (h-mrmr-GPR). Each node represents the target of an individual mrmr-GPR run, with the in-edges as relevant and non-redundant predictors (color of arrows indicates feature importance). The colors of edges refer to feature importance, and line width of nodes refers to R2 value for GPR surrogate model, see legend on the right.

DOI