I have a set of scores from two traits. After applying simple sum rule for fusion i.e. $f_{s}=\sum_{i=1}^{n} w_i s_i$, fused score is achieved. From the fused score, I need to evaluate EER (equal error rate). EER is a point where false accept rate (FAR) and false reject rate (FRR) holds equality.
How do I construct the objective function to minimize EER? How do i apply gradient descent to estimate optimal weights for my problem?