MultiTargetClassifier.java - A Multi-target Classifier.
To implement this interface, it is also necessary to extend MultilabelClassifier.
Implementing this interface only signals to the Evaluator that we are dealing with multi-target data,
and a different evaluation output is made. Training and classification is the same, using the
methods buildClassifier(Instances) and distributionForInstance(Instance), except that
the latter may return a vector of L*2 doubles instead of L. The extra values are probabalistic
information (i.e., the probability of class y[j] for the jth target is y[j*2]) that may be used by