(Combining multiple classifier systems) The following seven classifiers C1, C2, C3, C4, C5, C6, and C7 are used to classify an instance (subject, object, or data item) into four class labels A, B, C, and D with probability (5 points each).
Classifier C1 C2 C3 C4 C5 C6 C7
Confidence 0.63 0.56 0.7 0.9 0.4 0.85 0.75
Label 0.80 0.04 0.80 0.80 0.04 0.13 0.80
B 0.13 0.80 0.04 0.03 0.80 0.03 0.04
C D 0.03 0.13 0.03 0.13 0.03 0.04
0.13 0.04 0.03 0.13 0.04 0.13 0.04
0.03 0.13 0.04 0.13 0.80 0.03 0.13
(a) Using majority voting without confidence considerations, what is the final choice of the class label?
(b) Same as (a), but taking into consideration the confidence of each classifier?
What would be the final ranking for class labels A, B, C, D if these 7 classifiers are combined using average score combination?
Same as (c), but using average rank combination?
Same as (c), but using weighted score combination and confidence?
Same as (d), but using weighted rank combination and confidence?