7.5.2 Stopped Search

Stopped search refers to obtaining the network's parameters at some intermediate iteration during the training process and not at the final iteration as is normally done. This is, like the regularization, a way to limit the number of used parameters in the network. During the training the efficient number of parameters grows gradually and eventually becomes equal to the nominal number of parameters at the minimum of the MSE. Using validation data, it is possible to identify an intermediate iteration where the parameter values yield a minimum MSE. At the end of the training process the parameter values at this minimum are the ones used in the delivered network model.

In the following example, the performance of this stopped search technique is compared to that of a fully trained model.