A Model Based on Convolutional Neural Network for Online Transaction Fraud Detection

Algorithm 1

Find the best permutation of input features.

Input: The weight matrix , the rows of the weight matrix , initial state of the input

set , list of model accuracy rates , auxiliary sequence ( is the number of rows

below , satisfy the first condition ) and ( control the direction of

change)

Output: The best permutation of weight matrix.

1:

2:

3:

4: Access matrix

5: , ()

6: :

7: if    then

8:go to 19

9: end if

10: if    then

11:go to 14

12: end if

13: , , go to 4

14: if    then

15:input to model and calculate the accuracy

16:

17:if    then

18:return and that correspording

19:end if

20: else

21:

22: end if

23: , , go to 6