Enhanced Performance by Time-Frequency-Phase Feature for EEG-Based BCI Systems

Table 1

The classification results using different feature processing methods and different feature types.

	Power	IA	IP	IF	Power-IP	IA-IP	IA-IP-IF	Power-IA-IP-IF
No-scaled-MIFS
ELMs	0.69±0.02	0.71±0.04	0.78±0.03	0.71±0.03	0.71±0.02	0.74±0.05	0.74±0.03	0.69±0.02
SVMs	0.69±0.02	0.71±0.04	0.78±0.03	0.69±0.06	0.71±0.02	0.73±0.04	0.74±0.04	0.70±0.01
Scaled-MIFS
ELMs	0.71±0.03	0.71±0.03	0.77±0.04	0.76±0.03	0.78±0.03	0.79±0.03	0.82±0.03	0.82±0.03
SVMs	0.72±0.04	0.71±0.03	0.77±0.03	0.76±0.03	0.77±0.03	0.79±0.03	0.80±0.02	0.80±0.03
Scaled-mRMR
ELMs	0.69±0.04	0.71±0.03	0.83±0.03	0.80±0.04	0.85±0.02	0.85±0.03	0.90±0.03	0.91±0.03
SVMs	0.72±0.04	0.72±0.04	0.83±0.03	0.82±0.03	0.85±0.04	0.86±0.03	0.91±0.03	0.92±0.02