ACO-based hybrid classification system with feature subset selection and model parameters optimization

摘要: This work presents a novel hybrid ACO-based classifier model that combines ant colony optimization (ACO) and support vector machines (SVM) to improve classification accuracy with small appropriate feature subset. To simultaneously optimize the subset SVM kernel parameters, importance pheromones are used determine transition probability; weight of provided by both considered update pheromone. The experimental results indicate hybridized approach can correctly select discriminating input features also achieve high accuracy.