A Learning Framework of Nonparallel Hyperplanes Classifier

Table 3

Tenfold testing percentage test set accuracy (%) on binary-class UCI data sets for linear classifiers.

Data sets	TWSVM	SVM	GEPSVM	BHNPSVM
	Accuracy %	Accuracy %	Accuracy %	Accuracy %
	Time (s)	Time (s)	Time (s)	Time (s)
Hepatitis	82.89 ± 6.30^*	84.13 ± 5.58	80.07 ± 5.43	85.47 ± 1.36^*
	0.012	0.012	0.0006	0.0304
BUPA liver	66.40 ± 7.74^*	67.78 ± 5.51	61.33 ± 6.26	69.97 ± 0.56^*
	0.840	0.0549	0.0012	0.2143
Heart-statlog	84.44 ± 6.80	83.12 ± 5.41	75.37 ± 7.02	84.44 ± 0.56
	0.023	0.0281	0.0022	0.1092
Votes	95.85 ± 2.75	95.80 ± 2.65	91.93 ± 3.18	95.58 ± 2.75
	0.797	1.1446	0.0039	0.1027
WPBC	83.68 ± 5.73^*	83.30 ± 4.53	76.76 ± 6.67	81.32 ± 1.36^*
	0.012	0.0432	0.0002	0.0465
Sonar	77.00 ± 6.10	80.13 ± 5.43	73.16 ± 8.33	74.15 ± 1.73
	0.007	0.0946	0.0225	0.007
Australian	85.94 ± 5.84	88.51 ± 4.85	80.00 ± 3.99	85.27 ± 3.26
	0.3460	0.2350	0.0029	0.4250
Pima-Indian	73.80 ± 4.97^*	77.34 ± 4.37	75.47 ± 4.64	77.05 ± 0.48^*
	0.121	0.261	0.0016	0.4793
CMC	68.28 ± 2.21^*	67.82 ± 2.63	66.76 ± 2.98	77.86 ± 0.22^*
	1.247	0.597	0.0050	1.197
Mean accuracy	79.81	80.88	75.65	81.23
Mean time	0.38	0.27	0.004	0.29

A greater difference between BHNPSVM and TWSVM.