Abstract:
Based on a database of 153 alloys, a surrogate model was trained using machine learning approaches to predict compressive strain-to-fracture in high-entropy alloys. As part of the work, the accuracy of the impact of the architecture of a fully connected artificial neural network (the number of hidden layers and the number of neurons in hidden layers) on the prediction accuracy was evaluated. It was shown that with an increase in the number of hidden layers, the absolute error decreases - from 5.4
