Skip to main content

Table 3 Performance comparison using different architectures in CapsNet-SSP

From: CapsNet-SSP: multilane capsule network for predicting human saliva-secretory proteins

Architectures Accuracy Sensitivity Specificity Precision F-score MCC AUC
One-Lane Conv 0.792 (8.8e-08) 0.827 (5.0e-07) 0.758 (2.3e-07) 0.771 (4.0e-08) 0.811 (3.6e-07) 0.602 (1.6e-07) 0.863 (1.1e-06)
Multi-Lane Conv 0.812 (4.4e-08) 0.806 (6.4e-06) 0.818 (1.0e-07) 0.814 (1.0e-08) 0.810 (9.6e-07) 0.624 (4.1e-08) 0.869 (2.5e-06)
One-Lane CapsNet 0.832 (0.015) 0.847 (0.04) 0.818 (0.013) 0.822 (0.013) 0.834 (0.013) 0.665 (0.025) 0.915 (0.001)
Multi-Lane CapsNet 0.888
(N/A)
0.847
(N/A)
0.929
(N/A)
0.922
(N/A)
0.884
(N/A)
0.779
(N/A)
0.948
(N/A)
  1. The threshold is set where the MCC reaches the maximum value, and the values in brackets are p-values