Table 10 Ablation study results on RADIOML2018.01A (vCPU-16, 64 GB RAM).

From: Cloud-enabled automatic modulation classification using deep feature fusion and Moth-Flame Optimized ELM approach

Model Variant

Accuracy (%)

Sensitivity (%)

Specificity (%)

ELM (baseline)

78.14

75.62

76.30

MFO-ELM (no deep features)

84.89

81.42

83.13

InceptionV3 + MFO-ELM

90.34

87.65

88.72

ResNet50 + MFO-ELM

89.41

85.23

86.87

VGG16 + MFO-ELM

88.74

84.11

85.02

Feature Fusion + ELM (no MFO)

91.63

87.36

87.91

MFOP-ELM (proposed)

94.19

89.56

88.76

Back to article page