Particle Swarm Optimization Based Selective Ensemble of Online Sequential Extreme Learning Machine

Table 4

Comparison of algorithms for classification problems with sigmoid hidden nodes.

Datasets	Algorithm	Number of nodes	Number of networks	Training time (s)	RMSE or Accuracy		Testing dev.
New-thyroid	OS-ELM	20		0.0043	93.18%	89.66%	0.1138
	EOS-ELM	20	15	0.0627	94.32%	90.92%	0.0276
	SEOS-ELM (GASEN)	20	15	1.2476	95.14%	91.58%	0.0201
	SEOS-ELM (PSOSEN)	20	15	0.5012	95.23%	91.78%	0.0198
Image segmentation	OS-ELM	180		1.8432	97.07%	94.83%	0.0078
	EOS-ELM	180	20	36.2458	97.08%	94.79%	0.0055
	SEOS-ELM (GASEN)	180	20	432.1987	97.60%	95.12%	0.0045
	SEOS-ELM (PSOSEN)	180	20	254.0721	97.56%	95.21%	0.0043
Satellite image	OS-ELM	400		42.2503	92.82%	88.92%	0.0058
	EOS-ELM	400	20	853.2675	92.80%	89.05%	0.0026
	SEOS-ELM (GASEN)	400	20	8241.4093	93.54%	89.92%	0.0017
	SEOS-ELM (PSOSEN)	400	20	6928.0968	93.96%	90.16%	0.0018