Figure 2.
The effect of the horizon cut-off on the performance of Algorithm 1. We report the average significant ranks at using different values of . In (a–c), we show the ranks based on the area under the PR curve (AUPR), the area under the ROC curve (AUROC) and top-precision (TPR) respectively. These results are obtained using 40 small networks. The average top-precision significant ranks on a set of 40 large networks are shown in (d).