Table 5 BCI CIV 2b—Result of the mean and standard deviation of the accuracies of the experiments.

From: Motor imagery classification using sparse representations: an exploratory study

Exp. ID

S1

S2

S3

S4

S5

S6

S7

S8

S9

AVG

setup1 MLP

89.48 ± 1.91

89.12 ± 1.82

84.01 ± 2.4

96.19 ± 1.05

90.67 ± 1.9

87.8 ± 1.96

88.32 ± 2.0

83.75 ± 2.05

81.5 ± 1.93

87.87 ± 4.13

setup1 R.Forest

91.22 ± 1.65*

86.99 ± 1.79

83.08 ± 2.38

98.94 ± 0.52*

92.38 ± 1.35*

87.4 ± 1.93

89.86 ± 1.91*

88.03 ± 1.26*

83.69* ± 2.44

89.07 ± 4.55

setup1 SVM

89.11 ± 1.54

81.55 ± 1.48

75.71 ± 2.84

98.3 ± 0.76

86.03 ± 1.97

84.62 ± 1.83

88.75 ± 1.52

83.37 ± 2.14

79.43 ± 1.64

85.21 ± 6.15

setup1 SRC

88.17 ± 2.23

87.08 ± 1.77

85.64 ± 2.43

88.56 ± 1.97

86.13 ± 2.36

85.69 ± 2.62

91.03 ± 1.87

78.18 ± 2.38

72.46 ± 2.07

84.77 ± 5.47

setup1 SRMLP

86.7 ± 2.1

86.81 ± 1.84

86.51 ± 2.39*

94.75 ± 1.4

91.23 ± 1.57

87.19 ± 2.39

87.81 ± 1.98

86.78 ± 3.04

80.3 ± 3.11

87.56 ± 3.67

setup1 SRR.Forest

88.01 ± 2.0

87.37 ± 1.7

82.08 ± 2.62

95.18 ± 1.39

89.28 ± 2.04

89.51 ± 1.85

88.29 ± 2.05

84.51 ± 1.9

79.67 ± 2.24

87.1 ± 4.29

setup1 SRSVM

87.83 ± 2.63

88.07 ± 1.82

81.47 ± 2.3

94.47 ± 1.41

89.33 ± 1.49

88.9 ± 2.33

88.18 ± 1.74

84.47 ± 2.77

79.49 ± 2.61

86.91 ± 4.23

setup3 MLP

90.31 ± 1.63

88.01 ± 2.09

82.12 ± 2.91

96.36 ± 1.45

90.38 ± 2.08

87.68 ± 2.57

88.03 ± 2.24

80.8 ± 1.28

81.71 ± 2.27

87.27 ± 4.74

setup3 R.Forest

90.18 ± 1.72

88.74 ± 1.65*

84.04 ± 2.32*

98.13 ± 0.54*

92.72 ± 1.49*

87.37 ± 1.2*

88.86 ± 1.64*

87.27 ± 1.31*

79.58 ± 1.41*

88.54 ± 4.9

setup3 SVM

86.18 ± 1.46

81.17 ± 1.56

74.01 ± 1.91

97.07 ± 0.81

85.38 ± 1.46

85.08 ± 1.25

86.64 ± 1.34

82.73 ± 0.91

78.2 ± 1.12

84.05 ± 6.04

setup3 SRC

88.81 ± 1.74

85.26 ± 1.78

84.23 ± 2.21

64.23 ± 1.22

73.77 ± 1.82

72.6 ± 2.56

81.54 ± 2.35

72.12 ± 2.61

64.29 ± 2.41

76.32 ± 8.53

setup3 SRMLP

90.89 ± 2.37

85.12 ± 1.47

76.92 ± 1.6

75.2 ± 2.15

84.77 ± 1.97

82.91 ± 1.84

85.87 ± 2.39

81.14 ± 1.76

73.87 ± 2.61

81.85 ± 5.28

setup3 SRR.Forest

89.42 ± 4.87

84.12 ± 1.82

77.18 ± 6.83

76.1 ± 1.42

85.62 ± 2.33

82.87 ± 3.7

84.9 ± 4.2

81.33 ± 2.25

78.38 ± 3.3

82.21 ± 4.12

setup3 SRSVM

89.69 ± 5.35

83.86 ± 1.59

78.72 ± 7.9

76.12 ± 1.93

86.18 ± 2.29

83.61 ± 3.17

84.5 ± 4.57

80.95 ± 1.9

77.72 ± 2.92

82.37 ± 4.12

setup6 MLP

87.24 ± 2.31

83.21 ± 2.65

83.07 ± 2.61

97.69 ± 1.19

95.42 ± 1.75

89.32 ± 1.93

85.5 ± 2.62

92.14 ± 2.13

89.66 ± 2.21

89.25 ± 4.84

setup6 R.Forest

94.52* ± 2.03

95.63* ± 1.77

95.07* ± 1.91

98.67* ± 0.83

97.14* ± 1.08

94.38* ± 1.87

91.9* ± 2.38

95.31* ± 1.57

94.36* ± 2.16

95.22 ± 1.79

setup6 SVM

89.78 ± 2.15

87.79 ± 2.96

86.51 ± 2.71

97.19 ± 1.31

93.73 ± 1.98

91.87 ± 2.4

88.13 ± 2.49

91.29 ± 2.09

89.16 ± 1.6

90.61 ± 3.14

setup6 SRC

86.16 ± 3.02

85.16 ± 2.81

86.7 ± 2.59

94.63 ± 1.36

91.28 ± 2.2

88.0 ± 2.39

84.31 ± 2.22

86.17 ± 2.7

87.21 ± 2.29

87.74 ± 3.07

setup6 SRMLP

88.93 ± 2.04

82.62 ± 2.23

81.82 ± 2.75

96.79 ± 0.94

96.04 ± 1.38

91.02 ± 2.41

88.06 ± 2.08

94.23 ± 2.16

89.38 ± 2.57

89.88 ± 5.03

setup6 SRR.Forest

88.36 ± 2.85

85.02 ± 2.55

85.48 ± 2.94

97.11 ± 1.55

96.02 ± 1.16

89.7 ± 2.79

88.92 ± 2.99

93.06 ± 1.39

90.38 ± 2.02

90.45 ± 4.0

setup6 SRSVM

90.48 ± 2.34

88.3 ± 2.52

87.69 ± 3.0

97.14 ± 1.12

95.3 ± 1.65

91.34 ± 1.97

88.48 ± 2.38

92.38 ± 1.71

90.74 ± 2.18

91.32 ± 3.02

  1. For each setup, 3\(\times\) cross validation (10-fold) was performed, totaling 30 scores. S1–S9 identifies the subjects of the base. The (*) mark the statistical difference for p value < 0.05 between the best approach with a conventional classifier (Random Forest) and the best approach that used sparse representation.
  2. Significant values are in bold.