Table 9 Analysis of ablation study based on the activation functions.

From: Modified ShuffleNet trained on gradient pattern and shape-based features for lung cancer classification with improved M-SegNet segmentation

Measures

Accuracy

Sensitivity

Specificity

Precision

F_measure

MCC

NPV

FPR

FNR

without RRB in SegNet

0.928

0.927

0.929

0.932

0.929

0.880

0.923

0.070

0.072

conventional relu

0.917

0.916

0.918

0.921

0.919

0.858

0.911

0.081

0.083

conv Prelu

0.922

0.921

0.923

0.926

0.923

0.868

0.917

0.076

0.078

Conventional softplus

0.935

0.934

0.935

0.939

0.937

0.874

0.929

0.064

0.065

Back to article page