ClusterTools Contents

ClusterTools User Guide



Semi-Supervised Classifier


 T Partially labeled dataset to be used for training.
 V Untrained classifier, default FNNC (Fast NN classifier).
 U Untrained clustering procedure, default MODECLUSTF([],6).

 W Trained classifier.
 LABE Estimated class labels of the entire dataset T.


The partially labeld dataset T should be a PRTools dataset with missing  labels. Missing string labels have to be set to the empty string. Missing  numeric labels have to be indicated by a NaN. Such a dataset can be  constructed by T = [A;X]; inwhich A is a fully labeled PRTools dataset

and X is a matrix of doubles consisting of feature vectors in the same
space as T
The dataset T, neglecting all labels, is first applied to the cluster
procedure U, which should be a PRTools fixed mapping. The resulting
cluster indices and the known labels of T are used to estimate all object
labels by CLUSTC. The resulting datasets is used for trainig the
untrained classifier V.

See also

datasets, mappings, cluste, clustf, clusth, clustk, clustkh, clustm, modeclust, modeclustf, meanshift, clustc, clustk, fnnc, labeld,

ClusterTools Contents

ClusterTools User Guide

This file has been automatically generated. If badly readable, use the help-command in Matlab.