mRMR-ABC: A Hybrid Gene Selection Algorithm for Cancer Classification Using Microarray Gene Expression Profiling

Table 11

The classification accuracy of the existing gene selection algorithms under comparison when combined with the SVM as a classifier for six microarray datasets. Numbers between parentheses denote the numbers of selected genes.

Algorithms	Colon	Leukemia1	Lung	SRBCT	Lymphoma	Leukemia2
mRMR-ABC	96.77 (15)	100 (14)	100 (8)	100 (10)	100 (5)	100 (20)
ABC	95.61 (20)	93.05 (14)	97.91 (8)	95.36 (10)	96.96 (5)	97.22 (20)
mRMR-GA	95.61 (83)	93.05 (51)	95.83 (62)	92.77 (74)	93.93 (43)	94.44 (57)
mRMR-PSO	93.55 (78)	95.83 (53)	94.79 (65)	93.97 (68)	96.96 (82)	95.83 (61)
PSO [51]	85.48 (20)	94.44 (23)
PSO [52]	87.01 (2000)	93.06 (7129)
mRMR-PSO [22]	90.32 (10)	100 (18)
GADP [27]					100 (6)
mRMR-GA [21]			100 (15)		95 (5)
ESVM [53]			95.75 (7)	98.75 (6)
MLHD-GA [40]			97.1 (10)	100 (11)	100 (6)	100 (9)
CFS-IBPSO [50]					100 (6)	98.57 (41)
GA [54]	93.55 (12)
mAnt [39]	91.5 (8)				100 (7)