Figure 2

Sensitivity performance on synthetic data. This figure shows the achieved sensitivities of BrierScoreMF (gray) vs linear SVM (white) on the synthetic data sets for m = 50 and n ∈ {50,100} and a varying number of classes k ∈ {2, 3, 4, 5, 10, 15, 20, 25}. For low k values, the SVM is better than the BrierScoreMF algorithm. However, for more than 10 classes, BrierScoreMF clearly outperforms multi-class linear SVM. These results were obtained by averaging 100 seeded comparisons.