Dynamic Learning Rate in Deep CNN Model for Metastasis Detection and Classification of Histopathology Images

Table 5

Cosine annealing (with restart) performance metric.

Learning rate	Accuracy	Precision	Recall	F1-score	AUC
Cosine LR ()
0.01-0.001	0.8045	0.8446	0.8045	0.7983	0.94
0.01-0.0001	0.9068	0.9078	0.9069	0.9067	0.96
0.001-0.005	0.8861	0.8896	0.8858	0.8858	0.96
Cosine LR ()
0.01-0.001	0.8361	0.8364	0.8361	0.8360	0.91
0.001-0.0001	0.8975	0.9020	0.8975	0.8971	0.97
0.001-0.005	0.8953	0.8955	0.8953	0.8953	0.96
Cosine LR ()
0.01-0.001	0.8526	0.8652	0.8526	0.8512	0.95
0.001-0.0001	0.8714	0.8789	0.8713	0.8708	0.96
0.001-0.006	0.9183	0.9205	0.9183	0.9181	0.98
Cosine LR ()
0.01-0.001	0.7866	0.8175	0.7866	0.7809	0.92
0.001-0.0001	0.9038	0.9072	0.9038	0.9036	0.97
0.001-0.006	0.68	0.7782	0.677	0.6433	0.88