Table 5.

The comparative performance of DenseNet-121 with Mish and ReLU using different optimizer methods.

Activation Function& Optimizer Methods	Accuracy %	Precision %	Recall %	F1 Measure %	Training Time
Nadam & ReLU	98.62	98.56	98.69	98.61	123 min 10 sec
Nadam & Mish	98.88	98.83	98.91	98.87	123 min 29 sec
Adamax & ReLU	95.69	95.65	95.9	95.68	118 min 27 sec
Adamax & Mish	96.38	96.32	96.54	96.37	118 min 39 sec
Adam & ReLU	95.09	95.02	95.22	95.08	119 min 40 sec
Adam & Mish	95.43	95.37	95.59	95.42	120 min 17 sec
Adadelta & ReLU	91.82	91.75	91.83	91.78	116 min 22 sec
Adadelta & Mish	92.34	92.27	92.34	92.3	116 min 47 sec
RMSprop & ReLU	89.15	89.12	89.05	89.08	114 min 29 sec
RMSprop & Mish	91.65	91.59	91.62	91.6	115 min 09 sec
Adagrad & ReLU	85.19	85.29	84.92	85.04	117 min 48 sec
Adagrad & Mish	88.03	88.07	87.85	87.93	118 min 05 sec
SGD & ReLU	81.06	81.12	80.74	80.86	123 min 45 sec
SGD & Mish	82.53	82.69	82.15	82.31	123 min 57 sec