Research on the Intrusion Detection Model of Underwater Sensor Networks

Algorithm 2

SMOTE algorithm pseudo code.

Algorithm 2 Oversampling algorithm SMOTE(T, N, k)

Input: T——Number of minority class samples

   N——Amount of SMOTE

   k——Number of nearest neighbors

Output: (N/100)T

1. numattrs——Number of attributes

2. Sample[ ][ ]——array for original minority class samples

3. newindex——keeps a count of number of synthetic samples generated, initialized to 0

4. Synthetic[ ][ ]——array for synthetic samples

5. if N < 100

6.   then Randomize the T minority class samples

7.   T = (N/100)T

8.   N = 100

9. endif

10. N = (int)(N/100)

11. for i = 1 to T do

12    Compute k nearest neighbors for i, and save the indices in the nnarray

13.    Populate(N, i, nnarray)

14. endfor

15. nnarray——Storing nearest neighbor arrays

16. while N != 0 do

17.    nn = random(1,k)

18.    for attr = 1 to numattrs do

19.      Compute: dif = Sample[nnarray[nn]][attr]- Sample[i][attr]

20.      Compute: gap = random(0, 1)

21.      Synthetic[newindex][attr] = Sample[i][attr] + gap dif

22.    endfor

23.    newindex + +

24.    N − −

25. endwhile