SPIDER The Spider Objects

Group object

    A=GROUP(I,[G]) returns a group object initialized with 
    a cell array of algorithms I and an optional grouping type G.  
    This is used to collect a group of algorithms.
    Examples:  group({svm knn c45}),
               get_mean(train(cv(group({svm knn })),train(toy)))
    group='all'           This parameter stores grouping type:
                          'one_for_each', 'all' or 'separate' (default='all').
                          When results of training and testing are
                          output by a (grouped) algorithm,
                          they are stored in the same group type given
                          here. This is important for objects like chain, 
                          and get_mean which do not deal with data objects
        GROUP TYPE            
        'all':               each item is passed independently and separately
                             to all training objects, e.g
                               d=gen(toy); a=group({knn svm});
                               train(a,group({d d},'separate'))
                             gives 4 outputs.
        'one_for_each':      the n^th data item is passed to the
                             n^th training object in an group object, e.g
                               d=gen(toy);g=group({d d},'one_for_each');
                               a=group({knn svm}); train(a,g)
                             gives 2 outputs.
         'all':              passes all the data objects to a
                             single training object, e.g 
                               d=gen(toy); a=group({knn svm}); r=train(a,d);
                             gives 1 ouput.
   [NOTE: It is also possible to use the transpose operator with
   group, e.g: r=group({ {svm svm} {knn knn} })' will give
   the same result as group({ {svm knn} {svm knn}  })]