Deep Extreme Learning Machine and Its Application in EEG Classification

Table 2

Performance comparison of DELM with ELM, KELM, and MLELM on UCI datasets.

Dataset	Algorithm	#	Training accuracy	Testing accuracy	Training time (s)	Testing time (s)
Ionosphere	Basic ELM	Average	0.9207 ± 0.0151	0.9342 ± 0.0222	0.0044 ± 0.0074	0.0013 ± 0.0048
		Best	0.9500	0.9735	—	—
	KELM	—	0.9900	0.9735	0.0064	0.0017
	ML-ELM	Average	0.9112 ± 0.0159	0.9447 ± 0.0216	0.0115 ± 0.0116	0.0014 ± 0.0050
		Best	0.9500	0.9801	—	—
	DELM	Average	0.9503 ± 0.0111	0.9474 ± 0.0292	0.0164 ± 0.0079	0.0045 ± 0.0064
		Best	0.9750	0.9934	—	—
Diabetes	Basic ELM	Average	0.7983 ± 0.0062	0.7725 ± 0.0129	0.0072 ± 0.0132	0.0034 ± 0.0098
		Best	0.8125	0.8021	—	—
	KELM	—	0.7899	0.7917	0.6818	0.0314
	ML-ELM	Average	0.7666 ± 0.0207	0.7522 ± 0.0324	0.0091 ± 0.0143	0.0044 ± 0.0109
		Best	0.7951	0.8177	—	—
	DELM	Average	0.7871 ± 0.0079	0.7580 ± 0.0422	0.0641 ± 0.0143	0.0112 ± 0.0086
		Best	0.8038	0.8229