Algorithm 1 GINI-Select algorithm

Input: feature subsets S1, S2, S3, S4, number of features to be selected k

Output: Optimal feature subset F

1: Initializes the optimal feature collection as F ←∅

2: D ←Initializes the feature subset to be selected as ∅

3: j ← 0

4: while j ≤k do

5:({S1[j]}=S2[j} =S3[j]} =S4[j]}}) then

6:F ← F ⋃{S1 [j}

7: else

8:D←D{S1[j]}

9: end if

10: end

11: gini_importance[{D}] =0

12: Build CART decision tree based on data set D

13: Calculate the Gini importance of each feature D[i] in D gini_importance[i]

14: Select the former k-len(D) feature with the greatest Gini importance as f

15: F ← F ⋃ f

16: return optimal subset F