Sample-Based Extreme Learning Machine with Missing Data

Table 2

Classification performance comparisons over different missing ratios on the UCI benchmark data sets. Performance includes mean aggregate accuracy ± standard deviation and training time (in seconds).


	S-ELM	ZF-ELM	MF-ELM

Breast cancer	0.7185 ± 0.0151	0.6570 ± 0.0334	0.6901 ± 0.0183
Breast cancer	0.3473	0.2776	0.2947

Diabetes	0.7375 ± 0.0263	0.6852 ± 0.0467	0.6957 ± 0.0448
Diabetes	0.9315	0.7257	0.7112

Flare solar	0.5635 ± 0.0217	0.5369 ± 0.0348	0.5454 ± 0.0441
Flare solar	2.938	2.108	2.272

German	0.7624 ± 0.0250	0.6984 ± 0.0301	0.7150 ± 0.0257
German	3.748	2.327	2.374

Splice	0.6704 ± 0.0602	0.5734 ± 0.0275	0.5916 ± 0.0273
Splice	10.83	11.26	9.273

Waveform	0.8287 ± 0.0236	0.7206 ± 0.0271	0.7456 ± 0.0314
Waveform	120.5	99.26	108.5

Mathematical Problems in Engineering

Sample-Based Extreme Learning Machine with Missing Data

Table 2