Velocity-based reinitialisation approach in Particle Swarm Optimisation for feature selection

Particle Swarm Optimisation (PSO) algorithm is known to be better than Genetic Algorithm (GA) as fewer operators are needed in its algorithm. However, it still has some weaknesses such as immature convergence; a condition whereby PSO tends to get trapped in a local optimum. This condition prevents them from being converged towards a better position. Various techniques have been proposed to tackle this problem by many means. This paper attempts to integrate several velocity-based reinitialisation (VBR) approaches in PSO for solving feature selection problem. Five benchmark datasets of various features dimension were used to implement the approaches. The results were analysed based on classifier performance and the selected number of features. The findings show that the proposed VBR is generally significantly better than the existing VBR approaches.