BAGC

Bag classifier for classifying sets of object instances

     [WBAG,WOBJ] = BAGC(A,OBJCLASSF,BAGINDEX,BAGCOMBC,BAGCLASSF,BAGLAB)
      WBAG = A*BAGC([],OBJCLASSF,BAGINDEX,BAGCOMBC,BAGCLASSF,BAGLAB)
      WBAG = A*BAGC(OBJCLASSF,BAGINDEX,BAGCOMBC,BAGCLASSF,BAGLAB)
      D = B*WBAG

Input
A	Training dataset with object labels and bag indices
B	Test Dataset with index list of bags, stored as label list
OBJCLASSF	Trained or untrained object classifier, default QDC
BAGINDEX	String or a label_list_index defining in which label list  of A the bag indices are stored, default 2.
BAGCOMBC	Combiner for objects in a bag, default VOTEC
BAGCLASSF	Untrained classifier for bags, default FISHERC
BAGLAB	String or a label_list_index defining in which label list  of A the bag labels are stored. Objects with the same  bag index should have the same bag label.  Default is the current labeling of A

Output
WBAG	Trained bag classifier
WOBJ	Trained object classifier
D	Classification matrix of bags in B

Description

This routine offers a classifier for bags (e.g. images) of objects  (e.g. pixels) stored in a single dataset. The objects in the training  set A should have at least two labels: bag labels (the class of their bag)  and bag indices, defining which objects belong to the same bag. These two  label sets should be stored by the ADDLABELS command in the dataset A.  Refer to the multi-labeling system (see MULTI_LABELING) offered by  PRTools. The current object labels of A can be the bag labels, but may  also be different, e.g. true object labels.

BAGINDEX should be a label_list_name or a label_list_index defining the  label list used for storing the bag indices that refer to the bag an  object belongs to. The same label_list_name or label_list_index should be  used for defining the bags of the test objects in B.

All objects in A are used to train the object classifier OBJCLASSF if it  is untrained. The current object labels are used for that. Classification  results of the objects in the same bag are combined by BAGCOMBC, which  can be any of the fixed combiners MEANC, PRODC, PERC, MAXC, etcetera.  This results for every bag in a single confidence vector for the classes.

If an untrained bag classifier BAGCLASSF is supplied, the bag confidence  vectors are used to train a bag classifier.

New bags, organised in a dataset like B, with the proper bag indices per  object stored in a label list with the same name or label_list_index as  used in A, can be classified by the bag classifier WBAG.

If no bag classifier BAGCLASSF was defined during training, just the  results of the object classifier WOBJ are returned combined by BAGCOMBC over the objects in the same bag in B. In this case the final result is  identical to B*(A*WOBJ)*BAGCC([],BAGCOMBC), provided that A has class  labels and B is labeled by its bag indices.

See also

datasets, mappings, multi_labeling, bagcc, loso, dataset/addlabels, dataset/changelablist,