A Binary Pso-aco Hybrid Algorithm for Feature Subset Selection

Feature Selection is the process of selecting a subset of features available, allowing a certain objective function to be optimized, from the data containing noisy,irrelevant and redundant features. This paper presents a novel feature selection method that is based on hybridization of ACO with a binary PSO to obtain excellent properties of two algorithms by synthesizing them and aims at achieving similar or better results than PSObased feature selection and ACO-based feature selection. The fundamental idea of this hybrid approach is that the PSO is employed at the beginning of the searching process,as PSO has the ability to eplore the search space .The best feature subset having best gbest that minimizes the classification error is selected.The searching process is switched to the ACO algorithm; the best subsets found by PSO is the initial population of ACO. From the features of the gbest having m features each ant start with randomly produce m-p feature subset .ACO works as a local search,wherein,ants apply pheromone-guided mechanism to refine the positions found by particles in PSO stage.In PSACO ,a simple pheromone-guided mechanism of ACO is proposed to apply as local search.Thus ACO helps PSO process for rapidly and effectively attaining the optimal and near optimal solution(subset). Proposed algorithm is applied to a biometric feature selection problem of left index knuckle dataset.