Skip to main content
. 2024 Nov 6;22(4):591–606. doi: 10.1007/s12021-024-09694-2

Table 3.

Computation time to calculate 1283 saliency maps per participant and across 370 participants in the data perturbation test set. Results are reported for the GPUs and CPUs listed in Section "Convolutional Neural Networks"

GPU CPU
All (mm:ss) Per participant (s) All (mm:ss) Per participant (s)
Gradient 0:17 0.03 1:42 0.26
Input X Gradient 0:08 0.02 1:05 0.17
Masked Gradient 0:10 0.03 1:03 0.16
Guided Backprop 0:10 0.03 1:09 0.17
Guided GradCAM 0:15 0.03 1:19 0.20
Integrated Gradient 4:31 0.68 51:37 7.80
Gradient SHAP 0:29 0.07 4:20 0.65