Double-Criteria Active Learning for Multiclass Brain-Computer Interfaces

Table 1

Details of the datasets including the numbers of the corresponding features and samples.

Dataset	Number of			Percentage of initial labeled instances (%)	Percentage of initial unlabeled instances (%)	Percentage of test instances (%)
	Features	Instances	Classes
Liver	7	345	2	10	40	50
Diabetes	8	768	2	10	40	50
Wdbc	30	569	2	10	40	50
Twonorm	20	7400	2	1	49	50
Hayes-Roth	4	160	3	10	40	50
Iris	4	150	3	10	40	50
Wine	13	178	3	10	40	50
Segment	19	2310	7	10	40	50
Letter	16	20000	26	1	49	50