Steps 

1  Normalization of feature matrix 
2  Calculating the center of classes 
3  Calculating the Euclidean distance of the classes center 
4  Integrating the same classes 
5  Calculating correlation between the new classes in both levels 
6  Feature selection in both levels 
7  Apply SVM Algorithm to the classes of the level 1 
8  Apply SVM Algorithm to the classes of the level 2 
9  Final decision 
10  end 