Fig. 5From: Boosting for high-dimensional two-class predictionTest-set error for different classifiers and number of variables in the simulation setting with complex separation between the classes. The figure reports the average test set error for the simulation setup with the complex separation between the classes for different number of boosting iterations (from 1 to 500); upper panels: 1000 variables, lower panels: 10000 variables. See text for more detailsBack to article page