Recognition of Flexion and Extension Imagery Involving the Right and Left Arms Based on Deep Belief Network and Functional Near-Infrared Spectroscopy
Table 2
DBN model classification results (%) for test dataset from one subject trained with training dataset from other subjects.
MODLE
(h1, h2)
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
DBN_S1
(10, 30)
81.79
78.97
80.43
76.89
78.52
76.87
78.69
80.06
68.79
78.93
DBN_S2
(10, 10)
77.92
79.23
76.27
75.06
77.92
75.97
77.67
74.26
75.31
69.93
DBN_S3
(10, 40)
68.71
66.17
70.94
67.46
68.71
64.93
66.56
64.48
63.26
64.43
DBN_S4
(10, 10)
79.25
77.49
76.85
79.29
75.43
79.25
77.49
71.85
77.29
69.25
DBN_S5
(10, 50)
72.06
68.16
72.79
71.93
73.97
69.06
61.69
69.79
69.93
66.06
DBN_S6
(10, 30)
79.56
77.43
76.87
78.25
76.32
82.93
79.23
67.26
66.76
79.31
DBN_S7
(10, 10)
73.56
76.43
76.87
68.93
69.32
77.64
78.56
69.26
71.97
73.67
DBN_S8
(10, 30)
77.06
76.83
75.97
76.43
76.12
75.34
75.56
81.47
76.67
77.67
DBN_S9
(10, 50)
64.26
65.31
72.93
67.13
69.48
70.26
60.43
71.76
73.96
70.77
DBN_S10
(10, 30)
75.96
77.23
69.45
77.9
65.87
74.12
75.94
76.87
76.98
79.79
DBN_S1, DBN_S2, …, DBN_S10, respectively, represent DBN models trained using training sets S1, S2, …, S10, and (h1, h2) are the number of units of two hidden layers of DBN.
