Comparison of particle swarm optimization and backpropagation as training algorithms for neural networks | Synapse