Table 5 Comparative analysis of pre-trained models without frozen convolutional layers.

From: Design of a deep fusion model for early Parkinson’s disease prediction using handwritten image analysis

Model

Precision

Recall

F1-measure

MCC

AlexNet

91.1

91.0

90.8

90.5

DenseNet 201

96.5

96.2

96.3

96.1

SqueezeNet1.1

90.2

90.2

90.1

89.6

VGG-16 Net

88.2

88.0

87.9

87.3

VGG-19 Net

89.6

89.4

89.5

89.0

ResNet-50

94.4

94.1

94.2

93.8

ResNet-101

95.3

95.0

95.1

94.7

MobileNetV1

96.8

96.5

96.6

96.4

MobileNetV2

97.4

97.1

97.2

97.0

GoogleNet

92.21

94.67

94.67

94