Skip to main content


Table 3 Pairwise Wilcoxon tests on AUROCs of different methods

From: Predicting enhancers with deep convolutional neural networks

  gkmSVM 4conv2pool 4conv2pool4norm 4conv 6conv3pool 6conv3pool6norm
gkmSVM 5.1e-3 5.1e-3 5.1e-3 5.1e-3 5.1e-3
4conv2pool 4.6e-2 5.1e-3 5.1e-3 9.6e-1
4conv2pool4norm 5.1e-3 5.1e-3 2.8e-2
4conv 2.4e-1 5.1e-3
6conv3pool 6.9e-3
  1. We perform pairwise Wilcoxon tests on AUROCs of the six methods. Tests are conducted with the alternative hypothesis that the AUROCs of two methods are different in their medians. Small p-values indicate that two methods have different performance