Table 5 Quantified performances of DenseNet121 with FS and TL architectures.

CNN model	Architecture	K-fold 5	Metrics
CNN model	Architecture	K-fold 5	Loss	Accuracy	F₁-score	Precision	Recall	Specificity
DenseNet121 100 epochs Batch size = 32	FS	Fold 1	0.6835	0.7215	0.7104	0.7272	0.6959	0.8705
		Fold 2	0.6175	0.7166	0.6863	0.6916	0.6814	0.8627
		Fold 3	0.6203	0.7141	0.7093	0.7133	0.7055	0.8719
		Fold 4	0.6174	0.7200	0.7200	0.7284	0.7124	0.8840
		Fold 5	0.7234	0.7099	0.7061	0.7187	0.6949	0.8844
		Average	0.6524	0.7164	0.7064	0.7159	0.6980	0.8747
	TL	Fold 1	0.7780	0.8327	0.8193	0.8203	0.8185	0.9213
		Fold 2	0.6892	0.8227	0.7920	0.7920	0.7920	0.9112
		Fold 3	0.6635	0.8114	0.7804	0.7808	0.7800	0.9121
		Fold 4	0.7392	0.8162	0.8159	0.8169	0.8149	0.9320
		Fold 5	0.6757	0.8262	0.8242	0.8261	0.8224	0.9334
		Average	0.7091	0.8218	0.8064	0.8072	0.8056	0.9220

Quick links

Search