PRTools contents

PRTools manual

mclassc

MCLASSC

Computation of multi-class classifier from 2-class discriminants

   W = MCLASSC(A,CLASSF,MODE)

Input
 A Dataset
 CLASSF Untrained classifier
 MODE Type of handling multi-class problems (optional; default: 'single')

Output
 W Combined classifier

Description

For default MODE = 'single', the untrained classifier CLASSF is called to  compute C classifiers between each of the C classes in the dataset A and  the remaining C-1 classes. The result is stored into the combined  classifier W.

For MODE = 'multi' the untrained classifier CLASSF is trained between all  pairs of classes as well as between each class and all other classes.

This total set of C classes is combined by MINC. The use of soft labels
is supported.

 EXAMPLES   W = MCLASSC(GENDATM(100),QDC,'MULTI');

See also

datasets, mappings, minc,

PRTools contents

PRTools manual