An Extreme Learning Machine Based on Artificial Immune System

Table 5

Results of the five algorithms on the five benchmark regression datasets.

Dataset	Algorithm	Training	Testing Accuracy		Hidden
		Time (s)	Means	StDev	Nodes
Breast Cancer	AIS-ELM	14.898	2.46E-01	1.32E-02	30
	DS-ELM	15.367	2.53E-01	1.63E-02	30
	PSO-ELM	15.902	2.98E-01	2.35E-02	30
	SaE-ELM	16.342	2.65E-01	1.76E-02	30
	ELM	0.008	2.99E-01	2.07E-02	50
Parkinson	AIS-ELM	20.287	1.68E-01	3.35E-02	30
	DS-ELM	21.349	1.79E-01	3.67E-02	30
	PSO-ELM	22.547	1.87E-01	4.12E-02	30
	SaE-ELM	22.975	1.89E-01	3.95E-02	30
	ELM	0.011	2.12E-01	4.53E-02	50
Servo	AIS-ELM	14.942	8.81E-02	9.83E-03	20
	DS-ELM	15.256	9.41E-02	9.63E-03	20
	PSO-ELM	15.278	1.17E-01	1.35E-02	20
	SaE-ELM	15.456	1.07E-01	1.45E-02	20
	ELM	0.007	1.35E-01	1.95E-02	30
Yacht	AIS-ELM	13.478	1.76E-01	4.32E-02	20
	DS-ELM	13.755	1.87E-01	4.52E-02	20
	PSO-ELM	13.834	2.45E-01	5.63E-02	20
	SaE-ELM	14.292	2.23E-01	5.60E-02	20
	ELM	0.006	2.67E-01	8.43E-02	30
SinC	AIS-ELM	33.012	6.12E-03	3.54E-04	30
	DS-ELM	33.514	6.35E-03	4.32E-04	30
	PSO-ELM	33.095	7.46E-03	5.41E-04	30
	SaE-ELM	33.821	7.93E-03	5.76E-04	30
	ELM	0.013	8.01E-03	3.84E-04	50