An Application of Adaptive Genetic Algorithm Combining Monte Carlo Method

We show the feasibility of this algorithm in finding tool commonality with N=10. Actually, we have experimented on different values of N. The value of each parameter is shown in Table I. Specially, we fixed β to be 1.0 in this section. As shown in Table II and Fig. 3, the performance actually gets better when we repeated the model for more times. However, the improvement is not significant. With the increase of N, the number of true positive alarms is steady while the number of false positive alarms decreased a little bit. To be specific, when N inclines from 5 to 10, the number of false positive alarms declines from 3 to 2, which, in turn, results in a slight increase of F1 score. There is no variation when N changes from 10 to 15. The trend when N grows from 15 to 20 is similar to the trend from 5 to 10. Taking the cost of computation into consideration, we finally chose N to be 10.