“…Input: initialize feature weight ω k 0 = 1 K ; Metric M k 0 , k = 1, ..., K; β; η; 2: Initialisation: Generate P training sample pairs randomly: , where p = 1, ..., P and k = 1, ...K. 3: for p = 1 to P do for k = 1 to K do case if φ 1 holds true then 7:ξ = 1, update weight ω k,p by(10) and M k p by(12); update weight ω k,p using Equation(10)and M k p using Equation(12); Output: M k , k = 1, ..., K and ω k , k = 1, ...K;…”