Skip to main content

Table 9 Summary of results for identifiers with different neural network shapes and activation functions

From: Species determination using AI machine-learning algorithms: Hebeloma as a case study

   

Metrics (/100)

Class

CG

Activation

Top 1

Top 3

Top 5

MRR

F1m

Species

7

ReLU

75.0

95.4

98.8

85.2

72.4

Species

7

Mish

75.7

95.1

98.8

85.7

72.9

Species

7

ReLU/ReLU

72.7

94.9

97.9

84.1

72.2

Species

7

ReLU/Mish

75.1

95.8

98.1

85.4

72.9

Species

7

Mish/ReLU

74.4

95.8

98.2

84.8

71.6

Species

7

Mish/Mish

75.3

95.8

98.4

85.5

72.3

Species

8

ReLU

76.7

95.8

99.2

86.5

72.1

Species

8

Mish

76.5

95.3

99.1

86.2

72.7

Species

8

ReLU/ReLU

75.8

94.9

97.9

85.5

69.5

Species

8

ReLU/Mish

75.9

95.3

98.5

85.7

71.2

Species

8

Mish/ReLU

75.9

94.7

97.7

85.6

71.2

Species

8

Mish/Mish

75.2

95.3

98.1

85.4

70.0

Section

7

ReLU

94.9

99.3

99.9

97.2

82.8

Section

7

Mish

95.2

99.3

99.9

97.3

83.5

Section

7

ReLU/ReLU

95.1

99.1

99.7

97.2

81.2

Section

7

ReLU/Mish

94.9

99.4

99.7

97.2

83.1

Section

7

Mish/ReLU

94.9

99.0

100.0

97.1

80.8

Section

7

Mish/Mish

94.6

99.1

99.9

97.0

80.7

Section

8

ReLU

94.5

99.4

100.0

97.0

80.0

Section

8

Mish

94.1

99.6

100.0

96.9

78.3

Section

8

ReLU/ReLU

93.8

99.4

99.8

96.6

79.0

Section

8

ReLU/Mish

94.5

99.6

99.8

97.0

78.7

Section

8

Mish/ReLU

94.1

99.4

99.8

96.8

78.8

Section

8

Mish/Mish

94.5

99.6

99.8

97.0

78.3