On the hybrid of genetic algorithm and particle swarm optimization for evolving recurrent neural network

This work describes a new evolutionary system for evolving recurrent neural networks based on the hybrid of genetic algorithm (GA) and particle swarm optimization (PSO), called HGAPSO. In HGAPSO, individuals in a new generation are created, not only by crossover and mutation operation as in GA, but also by PSO. The objective of PSO is to mimic and incorporate the maturing phenomenon in nature into GA. To test the performance of HGAPSO, a fully connected recurrent network is designed and applied to a temporal sequence production problem. In simulations, the performance of HGAPSO is compared to both GA and PSO, demonstrating its superiority.