A Neural Network Design Method Based on Fuzzy Evolutionary Programming and Layer-wise Method

A fuzzy evolutionary programming method to design the feedforward neural network is proposed. By evolving a population of neurons instead of neural networks,the length of coding is decreased and computation pressure is greatly alleviated.At the same time, the method not only simplifies the computation of the fitness, but also decreases the complexity of the fitness space.The simulation results show that the premature convergence in evolutionary programming is restrained effectively, and the learning efficiency and convergence precision for the weights of the multi layer feedforward neural networks are improved greatly.These results also show that the proposed method can produce very compact artificial neural networks in comparison with other algorithms.