Evaluating Machine Learning Models for Predicting Hardness of AlCoCrCuFeNi High-Entropy Alloys

This study evaluates the predictive capabilities of various machine learning (ML) al-gorithms for estimating the hardness of AlCoCrCuFeNi high-entropy alloys (HEAs) based on their compositional variables. Among the ML methods explored, a back-propagation neural network (BPNN) model exhibited superior predictive accuracy compared to other algorithms, including Support Vector Machine (SVM), Stochastic Gradient Descent Regressor (SGDR), Bayesian Ridge (BR), Automatic Relevance De-termination Regression (ARDR), Passive-Aggressive Regressor (PAR), Theil-Sen Re-gressor (TSR), Linear Regression (LR), and Random Forest (RF). The BPNN model achieved excellent correlation coefficients (R²) of 99.54% and 96.39% for training (116 datasets) and testing (39 datasets), respectively. Validation of the BPNN model on an independent dataset (19 alloys) further confirmed its high predictive reliability. Addi-tionally, the developed BPNN model facilitated a comprehensive analysis of the indi-vidual effects of alloying elements on hardness, providing valuable metallurgical in-sights. This comparative evaluation highlights the potential of BPNN as an effective predictive tool for material scientists aiming to understand composition-property rela-tionships in HEAs.