Smoothing Regularization for Extreme Learning Machine

Table 3

Comparison of success rate in ELM, ELML1, and ELMSL0 algorithms on classification data sets.

Data set	Training accuracy			Testing accuracy			Average nodes
	(%)	(%)	(%)	(%)	(%)	(%)
	83.40	81.62	81.13	73.97	75.50	76.68	100	63.90	60.20
	90.79	89.91	89.17	84.88	85.57	86.50	100	63.40	59.50
	89.72	87.92	88.11	78.17	80.28	82.89	50	31.80	29.05
	100.00	100.00	100.00	77.94	78.53	82.50	500	315.90	310.10
	98.52	97.56	97.83	94.83	96.45	96.62	100	63.50	59.70
	96.56	96.00	96.56	72.08	70.67	69.36	300	189.60	186.30
	100.00	100.00	100.00	60.35	60.40	59.01	300	192.15	187.95
	98.90	99.25	98.80	91.80	92.40	96.00	50	30.70	27.60
	100.00	100.00	99.92	92.76	93.28	96.55	70	41.80	39.70
	98.47	96.53	96.67	84.92	80.17	88.20	60	35.80	33.75
	99.35	96.79	96.00	71.86	76.94	80.75	50	29.80	27.20
	92.78	91.27	90.89	81.35	84.60	86.10	60	35.55	32.60
	99.35	98.57	98.20	94.05	94.29	95.00	50	30.35	27.65
	90.21	83.36	82.37	49.62	58.75	60.57	200	125.85	121.20