The process of the network learning till the moment of its premature termination can be traced also on the training state plots shown in Figure 5 b, which document that the Levenberg–Marquardt gradient of the learning algorithm, as well as the network validation error mu, show a decreasing trend, with the number of iterations (epochs), in which the MSE of validation data has increased its expression as a parameter value fail, is zero.
← all excerpts
Prediction of the Tensile Response of Carbon Black Filled Rubber Blends by Artificial Neural Network.
1
—
—