... with the Bayesian Regularization Backpropagation Algorithm, in which the weights and biases are assumed to be random variables with specified distributions. This is needed to avoid the potential problem of overfitting the dataset, which is a common problem in machine learning algorithms trained with well-log data...