8.2 Examples

Following are examples using measured data from a DC motor and a hydraulic actuator. Two subsequent examples show how you can handle the bias-variance problem in two different ways. The first way is to use fewer parameters in your neural network model. The second possibility is to modify the model structure so that a more appropriate model may be obtained. These are, therefore, alternatives to the approach discussed in Section 7.5, Regularization and Stopped Search.