Gene Selection via a New Hybrid Ant Colony Optimization Algorithm for Cancer Classification in High-Dimensional Data

Table 4

Parameters used for experiments (common parameters for MWIS-ACO-LS).

Common parameters for MWIS		Value
	The correlation for crating a vertex	0.35
Common parameters for ACO
m	Population size	30
n	The number of iteration	100
α	Influence of the pheromone	1.2
β	Heuristic information	0.2
	Local evaporation of pheromone	0.002
	Local evaporation of pheromone	0.06
λ	Factor of updating pheromone	1.6
	The initial pheromone of pathway 0	1.0
	The initial pheromone of pathway 1	1.0
	The lower pheromone	0.05
	The upper pheromone	1.4
	The weight coefficient in the fitness function	0.99
Common parameter for the local search algorithm
	The maximal number of iteration without improvement	5
	The maximal number of Add operations	3
	The maximal number of Del operations	5
Common parameter for the backward generation algorithm
	The maximal number of iteration without improvement	20
	The maximal number of Add operations	0
	The maximal number of Del operations	2
Common parameter for the KNN classifier
K	The number of neighbors	1
The distance used	The Euclidean distance
Common parameters for the logistic regression model (LR)
	The regularization term constant	0.0001
	The weight given to into the Elasticnet	0.5