Algorithm 3: Features selection using ReliefF method
Input: M learning instances with L features and C classes
Output: the vector $w=W\left[f_1,f_2,f_3,\dots ,f_l\right]$
Step 1	[Initialized] Set all weights W (f_l) = 0, where 1,2,3…,l
Step 2	for i = 1 to m do
Step 3	Randomly select an instance $R_i$
Step 4	Find k nearest histograms $h_j$ of $R_i$
Step 5	Repeat for each class c≠ class ($R_i)$ do
Step 6	From class c, find k nearest misses $m_j$(c) of $R_i$
	[End for loop]
Step 7	For l = 1 to L, do
Update W(F) by using Equation (1)
	[end for loop]
	[end for loop]
Step 8	[end]