Table 3 Performance of 1D CNN with two convolution layers according to varying number of channels of the second convolution layer from 8 to 128 with the optimized first convolution layer with 64 channels.

From: Electrospray mode discrimination with current signal using deep convolutional neural network and class activation map

Architecture

Test accuracy (%)

Test precision (%)

Test recall (%)

Test F1-score (%)

Conv1D(64) + Conv1D(8)

94.44

96.30

90

92.16

Conv1D(64) + Conv1D(16)

96.30

97.08

93.33

94.80

Conv1D(64) + Conv1D(32)

96.30

97.44

93.33

94.96

Conv1D(64) + Conv1D(64)

94.44

96.30

91.67

93.48

Conv1D(64) + Conv1D(128)

94.44

95.75

91.94

93.44

Back to article page