Application of Deep Learning in Neuroradiology: Brain Haemorrhage Classification Using Transfer Learning

Table 2

Models learning parameters.

	CNN	AlexNet	AlexNet-SVM
Learning parameters	Values	Values	Values
Training ratio (%)	80	80	80
Initial learning rates	0.001	0.01	0.01
Number of epochs	100	200	140
Training accuracy (%)	92.89	94.12	96.34
Testing accuracy (%)	90.65	92.13	93.48
Achieved mean square error (MSE)	0.092	0.087	0.054