Algorithm 2. ReliefF algorithm

1:Input: Features: F, time of iteration: m, the number of features: a

2:Initialize the weight value W corresponding to all features to 0 and T to the empty set;

3:fori = 1 to m, do

4:Obtain the sample point $S_i$ randomly from the training data set $S_1,S_2,S_3\dots \dots S_n$;

5:Calculate k nearest neighbor samples $S_{iN{H}_h}^{}(h=1,2\dots k)$ from the training data set of the same type as $S_i$; calculate the k nearest neighbor samples $S_{\mathrm{i}N{M}_h}^{}(h=1,2\dots k)$ from each training data set different from $S_i$;

6:Forj = 1 to a, do

Equation (5),

7:end for

8:end for

9:Forj = 1 to a, do

10:If$\mathrm{W}(F_j)\ge \delta$

11:Add feature $F_j$ to T

12:end if

13:end for

14:Output: selected feature subset T