Entropy-Based Pattern Learning Based on Singular Spectrum Analysis Components for Assessment of Physiological Signals

Table 11

Comparison results of the identification accuracies for the proposed entropy-based pattern learning to identify specific biosignal patterns from EEG and EMG signals with SSA components representation by the use of different window lengths.

Eye states identification from EEG signals	Window length	6	10	14	16	20	30
	DimEnM	6	10	14	16	20	30
	DimIF	4	8	10	11	16	22
	Accuracy (%)	90.00	96.50	93.50	88.00	88.00	90.00
	EnM	FuzzyEn
	PatterC	SVM
	CrossVS	LOOCV
Physical actions classification from EMG signals	Window length	5	10	15	20	25	30
	DimEnM	5	10	15	20	25	30
	DimIF	5	10	15	20	24	30
	Accuracy (%)	72.50	81.25	75.00	87.50	70.00	75.00
	EnM	FuzzyEn
	PatterC	SVM
	CrossVS	LOOCV

DimEnM, dimension of entropy measures; DimIF, dimension of input features; EnM, entropy measures; PatternC, pattern classifier; CrossVS, cross-validation strategy; SVM, support vector machine; LOOCV, leave-one-out cross-validation.