A Novel Approach to Ensemble Classifiers: FsBoost-Based Subspace Method

Table 8

Level 1 classification results for the B dataset.

k-nearest neighbor classification algorithm
NP	k = 2			k = 2			k = 4
	DF
	Euclidean			Euclidean			Euclidean
FS	0			1			2
NF	68			25			15
Class	Sen	Spe	Acc	Sen	Spe	Acc	Sen	Spe	Acc
E	0.84	1.00	91.96	0.85	0.83	92.30	0.83	0.98	90.52
N-E	1.00	0.84		0.99	0.98		0.98	0.83
AUC	0.92			0.85			0.92
Kappa	0.84			0.92			0.85
F-measure	0.91			92.30			0.92
10-fold (%)	88.13			89.72			90.50
Probabilistic neural networks
NP	Spread			Spread			Spread
	0.11			0.21			0.31
FS	0			1			2
NF	68			25			15
Class	Sen	Spe	Acc	Sen	Spe	Acc	Sen	Spe	Acc
E	0.84	0.98	91.00	0.79	0.58	89.17	0.58	1.00	78.65
N-E	0.98	0.84		0.99	1.00		1.00	0.58
AUC	0.91			0.78			0.88
Kappa	0.82			0.88			0.76
F-measure	0.90			89.17			0.86
10-fold (%)	0.11			0.21			0.31
Support vector machines
NP	BoxConstraint
	4			15			3
FS	0			1			2
NF	68			25			15
Class	Sen	Spe	Acc	Sen	Spe	Acc	Sen	Spe	Acc
E	0.97	1.00	98.48	0.96	0.92	97.22	0.92	0.97	94.48
N-E	1.00	0.97		0.99	0.97		0.97	0.92
AUC	0.98			0.94			0.97
Kappa	0.97			0.97			0.95
F-measure	0.98			97.22			0.97
10-fold (%)	99.22			97.35			94.07

DF: distance function, Sen: sensitivity, Spe: specificity, Acc: accuracy (%), NP: network parameters, FS: feature selection, NF: number of features, EC: ensemble classifier, E: epilepsy, and N-E: nonepilepsy.