Fig. 7

Genetic algorithm evaluation. Left panel: evolution of fitness function F = (f _T , − N _ess. nodes , N _nodes , − N _edges ) and s _all score for the best network (i.e. with minimal fitness function) obtained during a run of the optimization method using the same input data as in Fig. 5. Blue boxplots summarize values obtained with 200 independent runs of the optimization method. Grey boxplots show corresponding results obtained with 200 “random runs”. Note that each box summarizes values obtained during 10 iterations (2000 data points). Right panel: fitness function and s _all score for the best network (according to the fitness function) obtained after a specified number of runs (horizontal axis). Each box summarizes 500 values, obtained by randomly sampling the given number of runs among 200 independent runs of the optimization method