Fig. 18

Results for ML on normal instances with the random forest model trained on each of the datasets given in Table 3, where a gives the results when the ML model is trained on normal data, and b gives the results when the model is trained on LGT data. For each training dataset, identified by the parameter pair \((\max L, M)\), the value shown in the heatmap is the average, within each instance group, of the reticulation number found by ML divided by the reference value. We used a group of 16 instances for each combination of parameters \(L \in \{20, 50, 100\}\) and \(R \in \{5, 6, 7\}\)