Performance results for the 3 datasets: the average performance (over subjects) in AUC (in ), the average percent of selected sensors (Sel), and the value of the Wilcoxon signed-rank test for the AUC when compared to the baseline SVM's one.
Methods
Datasets
EPFL dataset (8 Sub., 32 Ch.)
UAM dataset (30 Sub., 10 Ch.)
ErrP dataset (8 Sub., 32 Ch)
Avg. AUC
Avg. Sel
value
Avg. AUC
Avg. Sel
value
Avg. AUC
Avg. Sel
value
SVM
80.35
100.00
—
84.47
100.00
—
76.96
100.00
—
SVM-1
79.88
87.66
0.15
84.45
96.27
0.5577
68.84
45.85
0.3125
GSVM-2
80.53
78.24
0.31
84.94
88.77
0.0001
77.29
29.84
0.5469
GSVM-p
80.38
77.81
0.74
84.94
90.80
0.0001
76.84
37.18
0.7422
GSVM-a
79.01
26.60
0.01
84.12
45.07
0.1109
67.25
7.14
0.1484
Best performing algorithms for each performance measure are in bold.
The value refers to the one of a Wilcoxon signed-rank test with SVM as a baseline.