Information Technology Reference
In-Depth Information
Fig. 3.10. Experimental quantification—Representation by CCA
with a complex network, which is regularized by early stopping of the train-
ing process. The bootstrap is used to assess e ciently the variability of the
network and its error in relation to the data. Combined with early stopping,
it is used to monitor the training process by automatically optimizing the
number of training epochs, whilst providing the statistical characteristics of
the generalization error.
Search WWH ::




Custom Search