Algorithm 1. Pseudocode of ReliefF algorithm:

Inputs: Instance set S and the number of classes C Output: Weight vector w Step 1: For any feature fa, a = 1, 2, …, d, set the initial weight wa = 0 Step 2: for i = 1 to m do Randomly select xi from S; Select the k-nearest neighbors hj from the same class of x; Select the k-nearest neighbors mj(c) from different class from x for a = 1 to d do Update the weight by (9): $$w\left(a\right):=\frac{w\left(a\right)-{{\displaystyle \sum }}_{j=1}^k\frac{dist\left(a,x_i,h_j\right)}{m-k}+{{\displaystyle \sum }}_{c\ne class\left(x_i\right)}\left\{\frac{P\left(c\right)}{1-P\left[class\left(x_i\right)\right]}\times {{\displaystyle \sum }}_{j=1}^{k}dist\left[a,x_i,m_j\left(c\right)\right]\right\}}{\left(m-k\right)}$$ (9) End End