AdaBoost-based multiple SVM-RFE for classification of mammograms in DDSM

Table 5 Comparison of methods by maximum Az value using 8 features (Mass)

	T	MGH	WU	WFUSM	SHH
SVM		0.95821	0.97247	0.92252	0.97401
SVM-RFE		0.96218	0.97734	0.92252	0.97401
ENSEMBLE		0.72102	0.74859	0.67307	0.94292
JOIN (1)		0.77944	0.88187	0.79655	0.92650
JOIN (2)		0.72102	0.77365	0.79200	0.90262
JOIN (3)		0.72102	0.75484	0.79200	0.86857
JOIN (4)		0.72102	0.75484	0.75765	0.86861
JOIN (5)		0.72102	0.71136	0.67307	0.73745
MSVM-RFE (bootstrap)	5	0.95821	0.97247	0.92423	0.97401
	10	0.95821	0.97851	0.92288	0.97525
	15	0.95947	0.97457	0.92288	0.97401
	20	0.95947	0.97705	0.92315	0.97401
MSVM-RFE (boost)	5	0.95821	0.97247	0.92314	0.97401
	10	0.95821	0.97616	0.92426	0.97401
	15	0.95947	0.97247	0.92314	0.97401
	20	0.95947	0.97387	0.92314	0.97401

ISSN: 1472-6947