Figure 6
From: Towards an efficient validation of dynamical whole-brain models
Components of the cost function \(\Psi\)cost for the 2Dim case. (A–E) Boxplots illustrate the distributions of the values of the components indicated in the titles (see text for details). All values were normalized by the respective maxima (given as one quantity over all algorithms for a given component). The individual cost values are shown on the vertical axes while the corresponding algorithms (NMA2D, PSO2D, CMAES2D and BO2D) are indicated on the horizontal axes. (F–J) For each cost component, histograms show the relative number of recommendations obtained by a given optimization algorithm across all subjects. This quantity was calculated by counting the number of subjects for which the respective algorithm is considered recommendable, i.e. where it provides the lowest costs among all four investigated methods. The absolute value was then divided by the total cohort size of 105 subjects. As before, the methods are listed on the horizontal axes. On the vertical axes, it is indicated for which fraction of subjects a particular method is recommended. The figure was created with MATLAB R2018a (www.mathworks.com).
