Modeling with Neural Networks: Principles and Model Design Methodology - Neural Networks: Methodology and Applications - page 129

Fig. 2.16. Training without regularization: variation of the norm of the vector of
parameters as a function of the numbers of relevant and irrelevant documents in the
training set
Fig. 2.17. Training with regularization by early stopping: variation of the perfor-
mance of a linear classifier as a function of the numbers of relevant and irrelevant
documents in the training set