Automated classification of polyps using deep learning architectures and few-shot learning

BMC Medical Imaging

Table 11 Results on the validation data set considering different learning rates

Model	Learning rate				Val-acc	Training speed
	0.01	0.001	0.0001	0.00016		Min/Epoch	#Epochs
BiT-R152x4	\(\checkmark\)				0.7890	\(\approx\) 30	4
		\(\checkmark\)			0.8213	\(\approx\) 30	8
			\(\checkmark\)		0.8140	\(\approx\) 30	10
				\(\checkmark\)	0.8156	\(\approx\) 30	10
EfficientNet-B7	\(\checkmark\)				0.7903	\(\approx\) 5.7	6
		\(\checkmark\)			0.8212	\(\approx\) 5.7	10
			\(\checkmark\)		0.7924	\(\approx\) 5.7	30
				\(\checkmark\)	0.7969	\(\approx\) 5.7	28
Ours	\(\checkmark\)				0.4668	\(\approx\) 3	19
		\(\checkmark\)			0.5938	\(\approx\) 3	23
			\(\checkmark\)		0.8242	\(\approx\) 3	10
				\(\checkmark\)	0.8950	\(\approx\) 3	8

ISSN: 1471-2342