The raw training data are a stream set as defined in section
4.1.4 - that is, they are streams that are of the same
type. Recall that the training data is defined as a subset T of
for which the function 
is known. For example, in the
Auslan training example, the raw training data would consist of
several examples of the signs we hope to classify. Each training
example would be labelled with its class, i.e the result of applying
the function to the stream.