Table 5 Comparison of different local branch architectures.

From: Hybrid local-global representation learning with stochastic Gaussian classification for underwater acoustic target recognition

Local branch architecture

Accuracy

Precision

Recall

F1-Score

baseline

84.03

86.98

85.51

86.24

ResNet18

86.74

87.81

87.91

87.27

ResNet34

87.44

88.49

88.43

87.96

Custom architecture version 1

88.48

89.42

89.41

89.41

Custom architecture version 2

86.96

88.18

87.93

87.57

  1. Residual block configurations with varying channel sizes are compared in terms of classification performance. All values are reported in percentage (%).
  2. Bold values denotes the best performance for this metric.
Back to article page