#50 Regarding Validation

[Pascal Deschenes]

1. There is no mention anywhere in the Complete Guide
of the setValidationPatterns() method. This is crucial
otherwise the validation error will not be
valid...(training error == validation error).

2. The documentation mentions that we should be
creating the NeuralNetValidator within the
cycleTerminated() method rather then the errorChanged()
method. (P. 56, Section 4.4)

3. 9.4.2 is using the NeuralNetValidator object
strategy while 4.4 mention the 'obsolete' strategy ('3.
The Monitor of the cloned net is set to these values:
a. The totCycles parameter is set to 1. b. The
validation parameter is set to true.').