The Hybrid Feature Selection Algorithm Based on Maximum Minimum Backward Selection Search Strategy for Liver Tissue Pathological Image Classification

Pseudocode 1

The pseudocode of the framework of the hybrid feature selection algorithm based on MMBS search strategy.

The hybrid feature selection algorithm based on MMBS search strategy /

Input: train data , test data , () (the number of feature of th subset )

Output: feature subset

Initialize feature set as empty,

//Add feature

Calculate and threshold of each feature;

Calculate total number of feature .

Assuming the number of feature after adding is , initialize .

for

  if

   ; //add to feature subset

   ;

  end

end

Train classifier by the features in , and get a classification model; Classify the test data by the

model, and the classification accuracy is .

// Remove feature

   represents the features of feature subset after adding features.

  ;

  for

 for

   ;

   Train classifier by the features in tempS, and get a classification model; Classify

   the test data by the model, and the classification accuracy is .

   if

    ; //Remove feature

    ; //Update threshold of accuracy

   end

  end

  end