Fig. 13
From: Predicting the dissolution kinetics of silicate glasses by topology-informed machine learning
Dissolution rate predicted by “topology-informed” machine learning (Model IV) using a Artificial Neural Network (ANN) and b Gaussian Process Regression (GPR) as a function of the measured dissolution rate—wherein the dissolution data of Glasses A ((Na2O)0.25(Al2O3)x(SiO2)0.75–x, training set) are used as a training set to predict the dissolution kinetics of Glasses B ((Na2O)x(Al2O3)x(SiO2)1–2x, test set). c Distribution of the prediction error for the training (solid line) and test set (dash line) by using the ANN (black) and GPR models (blue), respectively. The results offered by polynomial regression are added for reference. The error is here defined as the difference between predicted and measured dissolution rates
