Research on the neural network based on an improved PSO algorithm

A modified particle swarm optimization is proposed for artificial neural network studying, in which a window is introduced to detect the change of environment. Through reinitializing the parameters, the modified particle swarm optimization can enhance the diversity of the particles and solve the problems in the determination of the convergence result by the distribution of the particles when they are initialized. The particle swarm optimization often converges prematurely in solving problems under complex background. Another parameter will be tuned according to the condition of the search process, and this tuning will balance the global search capability and the local search capability. Experiments were conducted to illustrate the effectiveness of this new algorithm, and the results showed that the modified particle swarm optimization is advantageous over the unmodified technique.