Effects of Data Augmentation with the BNNSMOTE Algorithm in Seizure Detection Using 1D-MobileNet

Table 7

Performance of different data sets with different IEDP algorithms.

Dataset	Imbalance ratio	Accuracy (%)						F1-score (%)
		Before	ROS	SMOTE	BLSMOTE	SVMSMOTE	BNNSMOTE	Before	ROS	SMOTE	BLSMOTE	SVMSMOTE	BNNSMOTE
CHB-MIT	1 : 30	98.95	99.11	97.86	98.91	98.48	99.40	85.70	87.77	79.98	85.98	84.29	91.90
Bonn	1 : 4	95.86	96.15	95.60	96.77	96.36	98.30	88.75	89.74	89.01	91.77	90.48	92.11