Fig. 1: Results of deep neural network training. | npj Materials Degradation

Fig. 1: Results of deep neural network training.

From: Deep learning framework for uncovering compositional and environmental contributions to pitting resistance in passivating alloys

Fig. 1

a Mean absolute error in pitting potential of network predictions on the training and testing datasets (termed as loss and validation loss respectively) in course of training history. b Plot of the DNN predictions over the test dataset after training. c Absolute prediction errors of pitting potential (both the mean and distribution in the form of 95% confidence limits) from the trained DNN model within each alloy class are plotted in black along the axis on the left. The statistical representation of each alloy class within the training and test dataset is represented by the bar chart in blue along the axis on the right.

Back to article page