fig1

Figure 1. Applications of ML in predicting phase formation of HEAs. (A) Schematic of the artificial neural network-based ML model; (B) Comparison of the sensitivity measures of the 13 design parameters based on the result of the ANN model. The figures are quoted with permission from Zhou et al.^[37]; (C) Prediction based on empirical descriptors and interpretable ML, and the workflow for obtaining robust descriptors using interpretable ML algorithms; (D) The best 2D descriptors for phase prediction and subfigures represent the result for four categories: amorphous and crystal; solid solution and intermetallic; single BCC or FCC and dual BCC & FCC phases; and FCC and BCC. The figures are quoted with permission from Zhao et al.^[38].