Classification Class Structure

• NumClassifySupervisedC gives normal classification behaviour and requires training using Learn with labelled sample feature vectors, ie NumVLDataSetC.
• NumClassifyUnsupervisedC performs classification training with unsupervised or unlabelled data and is normally referred to as clustering. This is trained simply with a list of feature vectors. The clusters can be initialised using Initialise and then iteratively clustered using Cluster. The clusters are automatically initialised at the first call to Cluster but are not initialised again until Initialise is specifically called so ythat the clusters can be updated progressively using multiple calls to Cluster with changing sample sets.
• NumClassifyFusionC is a base class for fusion classifiers which fuse the results of a list of classifiers to give a prediction of increased likelihood. This classifier can not be trained as a whole as the individual elements can be of different types. Each classifier must be trained individually before creation of the fusion classifier or using handles to the individual classifiers.