An evolved recurrent neural network and its application

An evolved recurrent neural network is proposed which automates the design of the network architecture and the connection weights using a new evolutionary learning algorithm. This new algorithm is based on a cooperative system of evolutionary algorithm (EA) and particle swarm optimization (PSO) for evolving recurrent neural network, and is thus called REAPSO. In REAPSO, the network architecture is adaptively adjusted by PSO, and then EA is employed to evolve the connection weights with this. network architecture, and this process is alternated until the best neural network is accepted or the maximum number of generations has been reached. In addition, the strategy of EAC and ET are proposed to maintain the behavioral link between a parent and its offspring, which improves the efficiency of evolving recurrent neural networks. The recurrent neural network is evolved by REAPSO are applied to a temporal sequence and the state estimation of continuous stirred tank reactor system. The performance of REAPSO is compared to TDRB, GA, PSO and HGAPSO in these recurrent networks design problems, demonstrating its superiority.