PRTools contents

PRTools manual

bagcc

BAGCC

Combining classifier for classifying bags of objects

     DBAG = BAGCC(DOBJ,COMBC)
     DBAG = DOBJ*BAGCC([],COMBC)

Input
 DOBJ Dataset, classification matrix, output of some base classifier
 COMBC Combiner, e.g. MAXC (default VOTEC)

Output
 DBAG Dataset, classification matrix for the bags in DOBJ

Description

This routine combines object classification results of bags of objects  stored in DOBJ. It is assumed that the current labels of DOBJ are bag  identifiers and defining objects belonging to the same bag. Objects of  the same bag are combined by COMBC into a single classification result  and returned by DBAG.

DBAG gets as many objects as there are bags defined for DOBJ. Effectively  the first object of every bag in DOBJ is replaced by the combined result  and other objects of that bag are deleted. DBAG has the same feature  labels (most likely the class names) as DOBJ and stores as object  identifiers the bag identifiers stored in the label list of DOBJ A possible multi-labeling definition of DOBJ is preserved.

This routine is called by BAGC where needed.

See also

datasets, bagc, multi-labeling,

PRTools contents

PRTools manual