Figure 5. Distributions of the parameters assessing the quality of network reconstruction performed using ANNs for the first simulation study.

Plots depict the distributions of FNR (A), FPR (B) and AUC (C) expressed as mean value and 95% confidence interval across 100 simulated networks as a function of the number of iterations of the gradient descent (Ntrain) for different values of the ratio between data samples and model coefficients to be estimated (K) and of the learning rate (LR) of ANN training.