A Core Set Based Large Vector-Angular Region and Margin Approach for Novelty Detection

Table 2

Average geometric mean accuracy and standard deviation on datasets.

Dataset	ν-SVM (%)	SVDD (%)	MAMC (%)	Core set based LARM (%)
Australian	47.88 ± 14.46	56.03 ± 9.21	35.11 ± 10.36	63.67 ± 9.12
B. authentication	97.25 ± 1.66	98.60 ± 0.95	95.34 ± 3.65	98.60 ± 2.30
B. Cancer	92.38 ± 1.43	95.25 ± 0.89	92.75 ± 2.43	93.61 ± 3.53
Cod-rna	45.63 ± 15.18	NULL	NULL	75.46 ± 8.79
Covtype	NULL	NULL	NULL	57.51 ± 8.27
Diabetic	42.91 ± 9.34	52.26 ± 9.01	50.14 ± 8.54	53.50 ± 8.90
Fourclass	77.23 ± 7.34	81.74 ± 11.27	64.31 ± 13.04	81.97 ± 8.16
Glass	95.12 ± 4.27	80.78 ± 5.71	94.56 ± 4.47	96.56 ± 3.52
Heart	42.01 ± 14.10	52.23 ± 5.80	51.50 ± 10.84	54.29 ± 8.32
H. valley	43.05 ± 9.25	42.00 ± 10.79	35.76 ± 19.45	44.16 ± 2.34
Ionosphere	46.44 ± 16.10	66.54 ± 9.81	36.07 ± 16.80	71.17 ± 14.91
L. disorders	50.21 ± 5.83	54.72 ± 7.49	45.80 ± 6.58	58.55 ± 3.14
MC	20.46 ± 3.84	67.36 ± 3.87	41.57 ± 7.12	62.94 ± 5.46
SDD	40.26 ± 9.63	25.93 ± 9.54	20.64 ± 21.66	45.67 ± 10.30
S. segmentation	NULL	NULL	NULL	95.91 ± 1.92
Sonar	52.27 ± 8.87	31.39 ± 6.99	55.79 ± 11.36	46.37 ± 9.04
Shuttle	91.65 ± 5.49	40.23 ± 21.59	NULL	92.88 ± 4.11
Svmguide1	82.17 ± 6.91	90.21 ± 6.88	89.95 ± 4.18	91.83 ± 2.69
Wilt	82.14 ± 9.29	56.95 ± 15.52	64.76 ± 3.81	84.95 ± 13.94
Wine	82.32 ± 4.57	86.00 ± 6.95	87.19 ± 4.32	87.62 ± 1.97