Training artificial neural networks with dynamic particle swarm optimisation

Particle swarm optimisation has been successfully applied to train artificial feedforward neural networks before, however, considered problems were assumed to be static. Such assumption does not hold for many real-world problems. This article investigates the applicability of dynamic particle swarm optimisation algorithms as neural network training algorithms for dynamic classification problems. The performance of dynamic particle swarm optimization is compared to back-propagation, and dynamic particle swarm optimisation is shown to be a viable training algorithm for dynamic classification problems.