Data-driven hardness prediction and analysis of {AlCoCrF eNiT iCu} system of High Entropy Alloys

In this work, Vicker’s hardness of the {AlCoCrF eNiT iCu} family of HEAs was predicted using artificial feed forward neural networks. The model was trained on 127 experimental data samples collected from the literature. The network prediction exhibits an average regression value of R = 0.9634, which is the highest in the literature of Machine Learning-based hardness prediction of high entropy alloys. Moreover, the trained model was used to extend the dataset by generating hypothetical samples of compositions that satisfy the definition of High Entropy Alloys. Using Pearson Coefficient of Correlation, the relationship between the chemical elements and the hardness was studied. According to the results, T i and Co influence the hardness the most with correlation coefficients of 0.4664 and −0.3625, respectively.