Figure 4

Demonstration of semi-supervised score based on the real data sets. Four comparisons based on Spearman's rank correlation ((A) and (D)), AUC 10% ((B) and (E)), and AUC1% ((C) and (F)). for the Prostate data set ((A), (B), and (C)), and for the Colon data set ((D), (E), and (F)). In each of the panels (A)-(F), the horizontal axis denotes the conditions, i.e., the numbers K of observed labels, where N is the total number of labeled samples in the original data sets. Blue, green, and red dots denote the compared scores, SAM, ODP-S, and ODP-SS, respectively, calculated based on 30 trials of random selection of the masked labels for each condition. Means of the 30 trials for each condition are plotted with lines in the colors of the corresponding dots.