Table 3 Comparison of 5-fold cross-validation results of CNN-FSRF and CNN-SVM on Yeast dataset.

From: Predicting Protein-Protein Interactions from Matrix-Based Protein Sequence Using Convolution Neural Network and Feature-Selective Rotation Forest

Test set

Accu.(%)

Sen.(%)

Spec. (%)

Prec.(%)

Fscore(%)

MCC(%)

AUC(%)

1

89.27

99.73

78.99

82.35

90.21

80.35

88.34

2

87.89

99.91

76.13

80.36

89.08

78.11

88.42

3

89.05

100.00

78.18

81.97

90.09

80.06

89.11

4

87.48

99.56

75.20

80.33

88.92

77.21

87.48

5

90.89

99.38

82.14

85.15

91.72

82.94

90.91

CNN-SVM Average

88.92

99.72

78.13

82.03

90.00

79.73

88.85

CNN-SVM Standard Deviation

1.34

0.25

2.71

1.97

1.12

2.22

1.29

CNN-FSRF Average

97.75

99.61

95.89

96.04

97.79

95.57

97.54

CNN-FSRF Standard Deviation

0.54

0.22

1.07

1.02

0.53

1.05

0.66

Back to article page