A Framework on Performance Analysis of Mathematical Model-Based Classifiers in Detection of Epileptic Seizure from EEG Signals with Efficient Feature Selection

Table 10

Comparison between different classifiers with Haar, dB4, and Sym 8 wavelet features with PSO feature selection.

Wavelets	Classifiers	Sensitivity	Specificity	Accuracy	F1 score	Error rate	G-mean
Haar	Linear regression	52	61	56.5	54.45026	43.5	56.55
	Nonlinear regression	64	55	59.5	61.24402	40.5	59.57134
	GMM	51	74	62.5	57.62712	37.5	63.12524
	K-NN	55	87	71	65.47619	29	73.01289
	SVM (linear)	56	53	54.5	55.17241	45.5	54.50389
	SVM (polynomial)	82	80	81	81.18812	19	81.01003
	SVM (RBF)	85	89	87	86.73469	13	87.04667
dB4	Linear regression	71	81	76	74.73684	24	76.21739
	Nonlinear regression	72	77	74.5	73.84615	25.5	74.55129
	GMM	55	78	66.5	62.14689	33.5	67.30243
	K-NN	87	82	84.5	84.87805	15.5	84.56879
	SVM (linear)	66	76	71	69.47368	29	71.18052
	SVM (polynomial)	63	66	64.5	63.95939	35.5	64.51159
	SVM (RBF)	63	93	78	74.11765	22	80.24002
Sym8	Linear regression	59	89	74	69.41176	26	75.96269
	Nonlinear regression	57	52	54.5	55.60976	45.5	54.51081
	GMM	64	51	57.5	60.0939	42.5	57.62039
	K-NN	54	52	53	53.46535	47	53.00117
	SVM (linear)	53	56	54.5	53.80711	45.5	54.50389
	SVM (polynomial)	55	59	57	56.12245	43	57.01053
	SVM (RBF)	98	98	98	98	2	98