BrainNet: Optimal Deep Learning Feature Fusion for Brain Tumor Classification

Table 2

Comparison of prediction accuracy of brain tumors.

Accuracies %
Classifiers	Org. ResNet101	Optimized		Feature fusion	%Age difference (org VS FF)
		PSO	DE
Fine tree	89.6	88.8	88	92.6	3.3%
Linear discriminant	96	95.7	95.6	95.7	−0.3%
Cubic SVM	96.7	96.7	96.6	96.7	0%
Boosted trees	92.8	92.1	92.3	92.5	−0.3%
Bagged trees	95.5	95.4	95.3	94.5	−1%
Subspace discriminant	95.8	95.5	95.4	95.4	−0.4%
Narrow neural network	95.7	95.5	95.4	93.9	−1.9%
Medium neural network	96	95.8	95.8	94.4	−1.7%
Wide neural network	96.1	95.9	96.1	95.4	−0.7%