Table 7 The influence of optimizers on VADGAT’s performance.

From: Triple dimensional psychology knowledge encouraging graph attention networks to exploit aspect-based sentiment analysis

Optimizer

Lap14

Twitter

Rest14

Rest15

Rest16

Acc(%)

F1(%)

Acc(%)

F1(%)

Acc(%)

F1(%)

Acc(%)

F1(%)

Acc(%)

F1(%)

adam

83.07

80.10

78.18

76.92

87.50

81.11

88.93

73.42

93.99

82.59

adadelta

53.92

31.57

47.98

25.77

65.00

26.26

59.23

28.62

76.30

29.43

adagrad

79.47

74.67

71.10

69.81

80.80

68.95

58.49

29.75

89.77

58.49

adamax

81.35

78.58

75.58

74.94

86.25

79.42

85.06

70.69

92.05

78.94

asgd

70.53

58.17

64.45

62.23

76.61

56.74

60.15

25.04

87.18

55.91

rmsprop

81.97

79.09

74.86

72.63

80.00

70.93

88.19

76.98

91.88

79.87

sgd

70.22

57.83

65.32

63.19

76.70

56.87

60.15

25.04

87.18

55.84

  1. The best result is in bold.